The Visual Display

Efiaard R.Tkfte 

The Visual Display 

of Quantitative Information 

SECOND EDITION 

Graphics Prex Cheshire, Connecticut

T 44$3 ex, 

[\ 

Coppight o 2oor by rdw:rd RolfTbfte 

?UBTISEED BY GRAPHTCS PRTSS ILC 

?osr OFFTCE Box 43o, CqtsgtRB, CoNNBcrrcur 06410 

rm nsha ro illtfitiotu dd b. Fed byldftd RolrTnftc. This worL n.y noi 6e @pieA EFodlccd, or hnihhn in vnoL or in 

I'd wirlDut wntu Fmi$id of i!. prujlnd, .cpt for bn€r*Gpts in cotucdion wilh Eviflj d lcnolIly :nt,5n. Us wirL ,,1 fod 

ofiDfo@arion *onse ra diml d@ic rlzpdid o. v!aEv*, cohpub sfMre, ft by sinnar u ;n{ihnr n€drodr now tnM 

edevdop€dinth€f!tueis&ridryfdn dd witnoft Eien Foni$ion '5ir!e !u61i5tE. A lubd of ilufatio4 e.epro&ccd !y 

Fni$iu; thF copyriglr+olds s d.diEd d Fge i9?. 

in the United States of Ansie4 Second edition, fowth yi ing,Januaryoo6

Contents 

PART I GRAPHICAL PRACTICE 

r Crcphiml Excellence 4 

z Gruphial Integriry J3 

3 So*tes oJ Ctaphical httegity afld Sophistiatiot 79 

PART II THEORY OF DATA GRAPHICS 

4 Data-Ink and Ctaphical Reiluigx 91 

5 Chattjunk: Vibrations, Crids, and Ducks 1o7 

6 Data-Ink Maxr ization and Craphical Design 721, 

7 Mtabijwctioning Graphical Elements r3g 

8 Dota Densiti and Suall Mubiples 161 

9 Aesthetics and Technique in Data Graphial Design 177 

Epikgue: Designs 

fu 

the Dkplay of Infornation r9r

Introdwction to the Second Edition 

This new edition provides high-resolution color reproductions of 

the many graphic.s ofWWifiam ?hyfair, adds color co other images 

*Lere appropriate, and includes all the changes a:rd correctiom 

zccumulated during the r7 printings ofthe fust edition. 

This book bega:r in 1975 when Dean Donald StoLes ofprinceton s 

Woo&ow Wilson School asLed me to teach statistics to a dozen 

i>umalists who llere vi)jiing rhrr ycar ro lerm some economjcr 

I annotated o collection ofre"dirgs. wirh r long seccion on 

*atistical graphics. The Jiterarue here was thin, roo often erimly 

devored to expl$ring use ofthe rulng oen and to p-..'E"ti"g 

"graphic 

standards" indiferent to the nature ofvisual evidence and 

qu:ntitative reasoning. Soon I wrote up some ideas. ThenJohr 

TuLey, the pherromenal Princeton statistician, suggested that we 

give a series ofjoinr seminars. Since the mid-r96os, Tukey h:d 

opened up the field, as his brilliant tecLnical contributiom made 

it dear that the study ofstatistical graphics was intellecturlly 

tspecuble nd notjusr abour pic chars rnd rLrlirg pens. 

After moving to Yale University, Ifnished the maauscript in 

1982. A publisher was interested but planned to print only z,ooo 

copies and to charge a very high price, contrary to my hopes for 

a vride readcrship. I also sought to design the book so as to make 

n selJ-exenplfying that is, the physical object itselfwould reflect 

tf,c intellcctual principles advanced in the book. ?ublishers seemed 

.ppalled at the prospect that an author might govcm design, 

Consequendy I investigated sclEpublishing. Tlns required a frstrrte 

book desigocr, a lot ofmoney (at least for a young professor), 

and a large garage. I fotmd Howard Gralh who had designed 

mauy museum catalogs with great care and cnft. He was willirig to 

work closely with this diiicult ruthor who was f,lled with all sorts 

ofopiniorx about design and typogmphy. We spent the summer in

his studio laying out the book, page by page. We were able to 

integrate graphics right into the text, son:retimes into the middle 

ofa sentence, eliminating the usual separation oftext and imageone 

ofthe ideas Zrsai Drsplay advocated. To fnance the book 

I took out another mongage on my home. The bank oflicer said 

this was the second most unusual loan that she had ever made; 6rst 

place belonged to a lo:n to a circus to buy an elephantl 

My view on selipublishirlg was to go all out, to maLe the best 

rnd most elegant and wonderful booL possible, withour compromise. 

Otherwise, why do it? 

Moit ofall, the book, as a thing in itsef gave to me fresh new 

eyes for the intellectual and aestheticjoy ofvisual evidence, visual 

reasoning, and visual understanding. 

Cheshire, Conneetiett

Introduction 

Data graphics visually display mcasured quantities by means of 

the conbined use ofpoinrs, lnres, a coordinare system, numbcrs, 

symbols, words, shading, and color. 

The use ofabstract, uon-reprcscntational pictures to show numbers 

is a surprisingly rccent invention, perhaps because of the divcrsity 

of skills required-rhe visual-anisdc, empirical-statistical, and 

mathematical. It was not until r75o-r8oo that sratistical graphics_ 

lengdr and area ro show quantity, time-series, scarterplots, and 

multivariate displays were invented, long after such triumphs of 

mathematical ingcnuity as logarithns, Cartesian coordinates, rLe 

calculus, and the basics ofprobabiLity theory. The remarkable 

William Playfair (1759 t84) developed or improved upon nearly 

:ll thc fundamental graphical designs, seeJ

The second part ofthe book provides a language for discussing 

graphics and a practical theory ofdata graphics. APplying to most 

visual displays of quantitative bformation, the theory l€ads to 

changc and improvements in design, suggests why some graphics 

might be better than othen, and generates new typer of graphics. 

TLe emphasis is on maximizing principles, empirical measures of 

graphical performance, and thc sequential improvement ofgraphics 

through revision and editing. Insights into graPhical design arc to 

be gained, I believe, &ord theories of whar males for excellencc 

in art, architecture, and prose. 

This is a book about the design of statistical graphics and, as such, 

it is concemed both with design and wirh statistics. But it is also 

about how to communicate infor:mation through the simultaneous 

presentation ofwords, numbers, and pictures. The design of statistical 

graphics is a universal rnatter-like mathemrticr-and is not 

tied to the unique Gatwcs of a particular language. The descriptive 

concep6 (a vocabulary for graphic$ and the principles advanced 

apply to most designs. I have at times provided evidence about fie 

scopc of these ideas, by showing how fiequently a pnnciple applics 

ro (a random sample of) news and scientfic graphics. 

Each year, the world over, somewhere between 9oo billion 

(9xro11) and u trillion (zlro12) images ofstatistical graphics arc 

printed. The principles of this booL apPly to most oftlose graphics. 

Some of thc suggested changes are small, but others are substantial, 

with consequcnces for hundreds of billions of printed peges 

But I hope also that the book has conseque.nces for the viewen and 

nukers of thot images-char they will never vies or creore 'tativ 

tical graphics the same way again. That is in patt because we are 

about to see, coliected here, so many wonderirl drawings, drose 

ofPlayfair, of Minard, of Marey, and, nowa&ys, of the computer. 

Most of all, then, this book is a celebration of data graphics.

Graphical Excellence 

Exccllence in statisrical graphics consists of complex idex 

communicated with clarity, prccision, and cficiency. Graphical 

displays should 

. show the data 

. induce the viewer to think about the substance rather than about 

methodology, graphic design, the tectrnology of graphic production, 

or something else 

. avoid distorting what the dara have to say 

, present many nunrben in a small space 

. make large data sets coherent 

. cncourage the eye to compare di{Ierent pieces ofdata 

. reveal thc data at several levels ofdetail, from a broad ovcrview 

to the 6ne structure 

. scrve a reasonably clear purpose: dcscription, exploration, 

tabulation, or decoration 

. be closeiy integrated with the statistical and verbal descriprions 

ofa data set. 

Graphics /eredl data. Indeed graphics can be more precise and 

revealing than conventional statistical cornputations. Consider 

Anscombe s quanet: all four ofthese data sets are described by 

exacdy the same linear model (at least until the residuals are exanined). 

10.0 

8.0 

13,0 

9.0 

11.0 

:14-0 

6_0 

12.0 

7.0 

5.0 

8_04 10,0 

6.95 8_0 

7.58 13_0 

8.81 9.0 

8.33 11,0 

9.96 14.0 

7,24 6.0 

4. 4.0 

'\0.84 72.0 

4.42 7,0 

5.68 5_0 

9.L4 

8.14 

4.74 

4.77 

9.26 

8.10 

6.13 

3.10 

9.13 

7.26 

10.0 

8.0 

13.0 

9.0 

11.0 

14.0 

6_0 

12.a 

7.0 

5.0 

7.46 

72.74 

7.L1 

7.41 

8.84 

6.08 

5.39 

8.15 

6.42 

5.73 

8.0 

8.0 

8.0 

8.0 

8.0 

6.58 

5.76 

7.71 

8.84 

4.47 

8.0 5,25 

19.0 12.50 

8.0 5.56 

8.0 7.91 

8_0 6.89 

N:11 

cquation of.egresioD line: Y : 3 + o5X 

standrd eno. ofestimte ofslolc : 0.118 

sun ofsquues X t:110.0 

rcgression suh of squa.es : 27.50 

rsidual sum of sq@res ofY : 13.7s 

corelatioD coediciot : .82

14 GRA?I]ICAL ?RACTICE 

Ard yet how they differ, as the grapLical display of the data 

makes vividly clear: 

Alrd likewise a graphic easily reveals point A, a wil&hot observation 

that will dominate standard statistical calculatiom. Note that 

point A hides in the marginal distribution but shows up x clearly 

exceptional in the bivariare scatter. 

F. J. Ansconbe, "cEpns h Statistical 

Aa lys," Amti.on Statittkiaf, z7 

Februrt r9j3),17 2t. 

stethe S. Brid md Stelhd !. Fidbss, 

"Re@t 

lconome.ric Modeling 

of Crime md P&isbDenr Suppoit for 

the Deten@@ Hypoth6is:" iD sElhd 

l. Fienbdg dd AlberJ. Reil3, Jr., cds., 

btdkataB oJ Cnac afld Cnnivl JrniE: 

Qrd/titatilc st di6 (washinston, D.C., 

1e8oJ, p. 89.

Of course, statistical graphics, just likc statistical calculations, are 

only as good as whar goes into them. An ill+pecifred or preposterous 

model or a puny dara scr cannot be rescued by a graphic 

(or by calculation), no matter how clever or fancy. A silly theory 

rneans a silly graphic: 

: -.004 

a 

€ -,* 

rlry au& sept, ocl Nd. 

so]A RrDAroN ND S]1h pRrG 

r1. New Yorl stftk pde (Bamt average). B. Sold R ctiarion, invqted, 

c- Lon.lon 3tocl pde, all by oonth!, 1929 (afrd cdia,Mara and 

u5 turn to the plactice ofgraphical cxcellence, rhe eficient 

mication of complex quantitative ideas. Excellcnce, nearly 

ofa muldvariare.orr. i" illLrsrrar.d herc lor frrndamenral 

and relational graphics. Thcse examples serve several 

providing a set ofhigh-qualiry graphics that can be 

d (and sometimes even redrawn) in constructing a theory 

graphics, helping to demorxtrate a descriptive terminology, 

eliry in briefabour the history ofgraphical developmenr. 

ofall, we will be able ro.eejuv hon good sratirtical 

b can be. 

CRA?HICALIXCXIIENCX 1J 

Edwlid R. Dewey and ndwin F. Dakin, 

cr.les: The s.ierc ofPrcnnian (New

16 GRA?HIcAT ?RACTICT 

Data Maps 

These six maps report the age-adjusted dcath rate from various 

typcs ofcancer for the 3,o56 couaties ofthe United States. Each 

map portrays some 21,ooo numbers.l Only a picture can carry such 

a volume of data in such a small spacc. Furthermore, all that data, 

dranLs to the graphic, can be thought about in many different 

ways ar rnany dili'erent levels ofanalysis-ranging from the contemplation 

of general overall parterns to thc dctcction ofvery 

ine county-by-county detail. To takejust a Gw examples, look 

at the 

. high death rates 6om catcer in the northeast part ofdle country 

and around the Great Lakes 

, low rates in an east-west band across the rniddle ofthe courtry 

' higher rates for men than for women in the soutL, padicularly 

Louisiana (cancers probably causcd by occupatioml exposure, 

lrorn working with asbestos in shipyard$ 

. unusual hot spots, including northern Min:resota and a few 

counties in Iowa and NebrasLr along the Mnsouri River 

. difi-erences in types ofcancer by region (for example, the lrigh 

rates of sromach cancer in the nonh-central part of the country 

probably the result of dre consumption of smoked Iish by 

Scandinavians) 

. rates in areas where you have lived. 

nre rnaps provide many lcads into drc causes-and avoidance 

of cancer. For example, the aurhors report: 

ln cerrain situations . . . the unusual orperience ofa county 

warrants fudrer investigation. For example, Salem County, 

NewJeney, leads the nation in bladdcr canccr mortality 

among whitc men. Wc anribute this exccs risk to occupational 

exposurcs, sincc about :5 percent ofthe employed persons in 

dns counry worL in thc chemical industry, particularly the 

manufacruring oforganic chemicals, wlnch rnay cause bladder 

tumors. After rhe frrding was communicated to NewJeney 

health oficials, a company in the area reported that at least 33o 

workers in a single plant had developed bladder canccr during 

the last 5o years. It is urgeDt that surveys ofcancer risk and 

programs in cancer control be initiated among worken and 

former worLers in this area.2 

ln,ch couty\ rate n loc.ted in tw. 

dnnensions dd, firrthei, at least fo6 

nuDbeB would be nece$r) ni re.onsdc 

he size ald sbape of.r.h cohty. 

This yields 7xl,016 entrics in. dru 

nra$ia sufi.i@t to rcproduce a nDp. 

Signi6candy hieh, but 

In highst de.ile, bui not 

Not signifE.ndy diferot 

Signiicanrly lower rhan 

,Robert Hoove., ThonlasJ. Manr, 

Innk w. McKay, od Joscph F. Fa! 

mmi,Jr., cncd br coDry: Ncw 

Resource for ntiologic Ches," S.ie,.o, 

189 (September 19, r97j), 1006. 

M^ps kom At16 oJ CM@t Moftalitr Jal 

U.s- Co!,r'6i r9Jei969, by Thonns J. 

MasoD, Ihnk W. Mchy, Roberr 

Hoover, wiluan J. Blot, dd Joeph F. 

Fnumdi,Jr. (wrshingron, D.c.: rublic 

Hmli\ Senice, Nadonal lNitutes of 

Hslth, ieTt. The six haps shoM hdc 

wete rcde$gled and ieilawn by 

Lawience Fahey and ldwdd Tufte. 

I 

I 

ffi 

n 

E

All rypes ofancer, white Gmales; 

age+djusted rate 6y county, 1g5o 1969 

AII types ofcancer, white males; 

age-rdjusted 

rate by coun.y, rssc-,+6s

Trachea, bronchus, and lung caacer; 

white femala; ageadjuted rate 

by county, 19Jo 1969 

Trachea, bronchuq aad lung cancer; 

white nales; age-adjusted rate 

by county, 19jo-1969

Stomach canccr, white f!ma1er; 

agc-adjustcd rate by county, 19jo 1969 

Sromach cancer, I'hite males; 

age-adjusted rate by counrl, 19Jo-1969

20 GR,{IEICAL IRACTICE 

The maps repay careful study. Notice how qui&ly and naturally 

our attention has been directed toward exploring the substantive 

content of the data rather than toward questions ofmethodology 

and technique. Nonerheles thc maps do have then flaws. They 

wrongly equate the visual importance of each county wirh irs 

geographic area rather than with rhe number ofpeople living in 

the counry (or the number ofcancer deaths). Our visual impression 

of the data is entengled wirh rhe circumsrance of geographic 

boundaries, shapes, and areas the chronic problem aflicting shadedin-area 

designs ofsuch "blot rnaps" or "patch maps." 

A fiuther shortcoming, a defect of data rather than graphical 

composition, is that the maps are founded on a suspect aleta source, 

dearh cerrificate repom on the cause of death. These reports fa1l 

under the influence ofdiagnostic fashions prevailing among doctors 

and coroners in particular places and times, a trcublesome 

adulterant of the evidence purporting to describe the already sometimes 

ambiguous mstter of the exact bodily site of the primary 

cancer. Thus part ofthe regional clustering seen on the maps, as 

well as some of the hot spots, may refect varying diagnostic 

cu.tom. and fads along wirh the acrual differ.nce' in canrer rare\ 

Data maps have a curious history. It was noi until the seventeenth 

century that the combination ofcatogmphic and statistical skills 

required to consrruct the data map cane together, fully 5,ooo years 

after the 6nt geographic naps were dram on clay tablets. And 

many highly sophisticated geographic maps were produced centuries 

before the frrst map contining any statistical mateial was 

drawn.3 For example, a detailed map with a full grid was engraved 

during the cleventh century A.D. in China. The Yn Chi Thu (Map 

ofthe Tracks of Yii the creat) shown here is described by Joseph 

Needham as the 

. . . most remarkable cartographic work ofits age in any 

culture, carved in stone in +rr37 but probably dating Gom 

before +rroo. The scale ofthe grid is 1oo l, to the division. 

The coastal outline is relatively lirm and the precision ofthe 

network ofriversystems extraordinary. The size ofthe oiginal, 

which is now in the Pei Lin Museum et Sian, is about 3 feet 

square. The name ofthe geognpheris not known. . . . Anyone 

who compares this map with the contemponry productions 

of turopern relgious co.mogrrphy crnnor bur be Jmr/ed ar 

the evrenr ro wluch Cline'e geography wrs at th:r ume a\e:d 

of rhe West. . . . There was nothing Like it in Europe till the 

Escorial MS. map ofabout +1JJo. . . ,4 

3Dab m?s are $uatly de$ribed as 

'lh€miic m.ps" in c,nogeph)'. For a 

thorough ,ccomr, sa Aitlff H. Robnsan, 

Eatl! Thenatn Mappikg in tlu 

Hitory a.f Cd asapb (Chi.aso,1982). 

On the history ofstatistical s.aphics, se 

H. cray FunLhotrsr. Histo.ical Develolmentof 

thecralhi€lRcpresdtation 

ofshtistical D,u," oJtn, 3 Novehbe( 

1917), 269-4o4i fd James R. Beniger 

md Dorothy r. Robyn, "Qunritarive 

Graphic in statislio: A BricfHistory," 

Andnan Statituian, 32 (Fetl@ry s73), 

.Josph Needhan, S.ie,a zd ci,t,iJd- 

I'or l, Crird (Cantridge, 19i9), vol. 3,

cat4 

iill 

cRA?ItIC,{l EXCEf f!NCt 21 

I. Chavdnes, L€s Deu Plus Anciens 

Spdcimms de la Catogralhic Chinoise," 

Brlletin de l *ole hon@ e de I'Exttne 

ot?tr,,3 Geoj). 1 lJ, Carr€ B.

22 GRATHlCA! PRACTICI 

Z"l 

e g 

t&D Ecce fornr ulam,vfum,arq ue 

fi rLrClunm T.rbularum Ptolonrri,cum quibufdam locis,r'n 

quibus ftudiofus Geographic fc fatis excrcetc poreft. 

L$'dt 

SEPTENTRIO. 

N,yy,h 

6"U 

7'flnt 

3' 

( It 

Oft4S tl'.. 

'1 ERID IE8. 

The rs+6 cdition of Co snograph;a by Pettrs Apianus contained 

examples ofmap design that show how very close EurcPean cartography 

by that time had come to achieving statistical graphicacy, 

even approaching the bivariate scatterplot. But, according to the 

historical record, no one had yet made the quantitative abstraction 

ofplacing r measured quantity on the map\ suface at the intersection 

ofthe two threads instead of the name ofa city, let alone 

rhe more diffrculr abstracdon ofrephcJng ladtude and longicude 

with some other dimensions, such as time and money. Indeed, it 

was Dot until 1786 that dre frrst cconomic time-series was ploned. 

0 

Ed

One of the fint data maps was Edmond Flalley's 16g6 chan 

showing trade winds and monsoons on a world map.s The detailed 

section below shows the cartographic symbolization; with, as 

Halley wrote, ". . . the sharp end ofeach litde snoaL pointing out 

that part ofthe Horizon, from whence the wind continuaily comesi 

and where Llere are Monsoons the rowr of.rroakr run alternarely 

backwa.ds and forwards. by wbich means rhey .rre rhicker ldenrerl 

than elsewhere." 

; ",fr a)\ TiI : $.P A 

-A'T'1 a- 

e-'.-r\- 

- c ,E'a!tr -' ' 

a 

"jit:'" i to 

,.2-.>-":.. 

't-/ a, 

':t:;t'{ 

-/ /- .a 

-7 

r,cf

24 GRA?HrCAt ?RACTICt 

An early and most worthy use ofa map to chart patterns of 

disease was thc famous dot map ofDr. John Snow, who plotted 

the location ofdeaths from cholera in cmtral London for September 

r8J4. Deaths were matked by dots and, in addition, the 

area's eieven water pumps were located by crosses. Examining the 

scatter over rhe suface ofthe map, Snow observed that cholera 

occurred almost entirely among those who lived near (and drank 

from) the Broad Street water pump. He had the handle ofthe 

contaminated pump rcmoved, ending the neighborhood epidemic 

which had taken more than 5oo lives.6 The pump is located at the 

center of the map, just to the right of the o in rRoAD srrxrr. Of 

course the link between the pump and the disease might have beea 

reveeled by computation and analysis without graphics, with some 

good luck and hard work. But, herc at leest, graphical ana\sis 

testifies about the date far more efiiciendy than calculation. 

Charles Joseph Minard gave quantity as well as directioi,r to the 

data measures located on the world map in his pomayal ofthe 

1864 exports ofFrench wine: 

6 l. w. cilterr, 

"Pioneer 

Maps of Health 

dd Disede in Eng1and," C@j'd?ltai 

Jd,bdl, D4G9j8),17>183. Shom h(c 

is a re&.wiDg ofjolD Snow's map. For 

a r.prodetion and &tailcd dalysis oftlc 

originrl na!, seendwrd Tufte, r,,tal 

Ertlandtio$: Lnass aad Qrtunn\, E iden@ 

fll Nd'drte (ch€sbne, com$icut, 

199?), chapr€r 2. Idealy, seJobn Snow, 

On the Mode oJ coM"niatiot o.f chaled

a 

rl 

ll 

tl 

l: 

i:! ; 

::. ; 

i ii : 

ji. i 

::: : 

;j. ; 

ii 

a: 

tii.:i 

i:r': t; :i:- ! 

ti:!: Ft: 

iii;: 

i;::: t:i:: 

i:":l 

j:::... 

"r^:n",,' 

'.rr, 

n- 

'/'r 

1. 

(.- 

! 

: l 

''1J 1, 

: : ! a 

i --< ;5.i 

i:=: ! 

-i ;: ;

26 GRAPIIICAL ?RACTICT 

Computerized cartography and nodern photographic rechniques 

have increased thc dcnsity of informrtion some 5,ooo-fold in dre 

bcst ofcurrcnt data maps compared to Halley's pioneering cffort. 

This map shows the distribution ofr.3 million galaxies (including 

some overlaps) in the northern galactic hcmisphcre- The map 

dividcs the sky into t,oz4xz,zzz rcctangles. The number ofgalaxies 

counted in each of the 2,273,328 rectangles is represented by 

ten gray tones; the darker the rone, dre greater the number of 

galaxies counted. The north galactic pole is rt the center. The 

sharp edge on thc left results from the earth blocking the view 

from the observatory. In the area near the perimeter of the map, 

the view is obscured by the interstellar dust of the gala-xy nr which 

we live (the Milky Way) as the line of sight passcs througL rhe 

flattened disk ofour gala:.y. The curious texture oflocal clusters 

ofg.laxe' seen in du. trul) ner vieq ofuhc univcrse wa. not 

anticipat€d by srudents ofgalaxies, rvho had, ofcourse, microscopically 

examined millions of photographs of galaxies before 

secing this macroscopic view. Although the clLrsters are clearly 

evidcnt (and accountcd for by a theory of galactic origin$, the 

seemingly random filamcnts may be happenstance. The producers 

of the rnap note the "stmng temptation to conclude that the galaxies 

are arranged in a rem:rkable {ilamentary pattem on scales 

ofapproxinmtely J' to 1J", but we caution drat this vnual impression 

may bc misleading beceu$ the eye tends to pick out linear 

parterns even in random noisc. Indecd, roughly sirnilar pttems 

rre seen on maps constructed ftom simulated catalogs whe'e no 

linerr structure has been built in. . . ."7 

The most extensive data maps, such as the cancer arlas and the 

count ofdrc galaxies, place millions ofbits ofinformation on a 

single page before our eyes. No other mcthod for the displiy of 

statistical information is so powerful. 

7 Michael Seldner,B.H. Sieles, ld$dd 

J. Grolh and P. Jancs E. Pccblcs, 

"New 

Reduction ofthe Lick Catalos of 

Ga]€iei Asttatunidl Jawidl, 82 C\ptil 

1977), 249-j r 4- See Gillian R. (napp, 

''Mining 

the Heavens: Thc SloanDigirrl 

Sky SuNey," Sll & T?leJ.or (Aqust 

ieeT), ao-a8i MarsaretJ. Geller andJohn 

P. Huchra, "Mapping the UniveBe. 

s&t A Tels.o?e (Ausust 199r, 1j1119.

clA?Htcal IxclrrIN c! 27

28 GRA?HIcAL ?RAcTIcT 

Tir[e-Series 

The time+eries plot is the most frequently used form ofgraphic 

design.s with one dimension marching along to the regular rhythm 

ofseconds, minutes, hours, days, wceks, nonrhs, ycars, centuries, or 

millennia, thc natural ordering ofthe tine scalc gives this dcsign a 

strcrgth and efiiciency of interpretation found in no other graphic 

arrangemeD!. 

This rcputcd tenth- (or posibly eleventh-) century illustration 

ofthe inclinations ofrhc planctary orbits as a function ofdme, 

apparentll parr of r text fbr monastery schools. is the oldesr known 

examplc ofan attcrnpt to show changing values griphically. It 

:rppears as a rnysterious and isolated wonder in the history ofdata 

g,ap\'' .. .rrcF rLe nexr e',r"nr graphc of " plotted Limc-crie. 

shows up some 8oo years later. According to Fulkhouser, the 

astronomcal contenr is confused and there are difricultics in rcconciling 

the graph and its accompanying texr with the actual movcments 

ofthe planets. Parricularly disconcerting n rhe wavy path 

ascribcd to thc sun.'An crasure and correction ofa curve occur 

..--. ,L^ -i,tJl^ ^f,t- ^--^l 

.ri4\t 

l \'PF 

!A random sam!]e of4,ooo g.rphcs 

dnm fron rj of tle sorld\ news 

!a!e6 d naBazins lublished fronr 

1974 Rr 1930 fou.d that more than 7t 

pcrccnt of all thc graphcs ptrblished 

wcrc tinc-serlca, Chaptcr 3 reporsnore 

,H. Gr,t lmkhouser, A Nore on a 

Tcnth Cedury Graph," Orin!, I 

(Ianudr 1916), 26e:62.

It was not unril rhe late 17oos thar dme-series charts began to 

appear in scienrific wrirings. Thrs drawing ofJohann Hcinrich 

Lambert, one ofa iong scries, shows the periodic variation in soil 

tempcrature in relation ro the dcprh undcr rhe surface. The greater 

the deprh, the greater the rime-1ag in temperature responsivencs. 

Modern graphc dcsigns showing time-series periodicities diFer 

little from those oflambert, although the data bases are far larger. 

a 

This plot of radio emissions from Jupiter is based on data collected 

by Voyagcr 2 in its pass ciose by the planct in July 1979. The rrdio 

inrensir;' increases and decrcases in a ren-hour cycle as Jupirer 

rolates. Maximum intensity occurs when rLeJovian north magneric 

polc is tipped roward fie spacecraft, indicating a norrhe 

hemisphere source. A sourhern source was dctected on July 7, as 

dre spacecraft neared the equatorial plane. The horizontal scale 

shows the disunce of the spacecraft from the planet measured in 

terms ofJupiter radii (R). Note the use ofdurl labels on the horizontal 

to indicatc both the d:te and disr:nce fromJupirer. The 

entire bottom panel also serves to label the horizontal scale, 

describing the changing orientation ofthe spacecrafr relative to 

Jupiter as the planet is approached. The multiple time--serics 

erforce not only comparisons witLin each series over rime (as do 

all time-series plots) bul also comparisons between the tluee 

diferent sampled radio bands shown. Thn ricHy multivariare 

display is based on 453,6oo instrumenr samples ofcight bits cach. 

The resulting 3.6 million bits were reduced by pcak and average 

p(ocessing ro rhe 18,9oo points acually plotted on thc graphic. 

GRA?IIICAL IXCXTl!NCT 29 

J. H. I-anberr, 4'o'err -l3ertin, r7?9). 

D. A. clrneti, W. S. (u.th, and F. L. 

sc2ri "Plasnx wave obs.rvarioDs Ncar 

Jupite' Ihirial Rcsulrs f.om voy4er 

2, s.tdra:06 (Novchb{ 2t, 19?e), 

987 991; md Ietter from Dorold A. 

Guhett to Edward R. Tufre, Jue 27,

30 GRA?HrCAl ?RACTTCI 

Time+eries displays are at their best fbr big data sets with real 

variability. Why waste the power of data graphics ol simple linear 

changes, 

which can usually be better summarized in one or two numbers? 

Instead, graphics should be reserwed for the richer, more complex, 

more difiicult statistical material. This New York City wearher 

summary for r98o depicts 1,888 numbers. The daily high and 

low temperatures are shown i:r rclation to the long-run average. 

Thc path of tle normal temperatures also providcs a forccast of 

expccted change over the year; in thc rniddlc ofFcbruary, for 

instancc, New YorL City residents can look forward to warming 

ar the rete ofabour 1.J degrees pcr wcek all the way to Ju1y, 

the yearly peak. This distinguished graphic succesfully organizes 

a large collection ofnumbers, makcs compansons between di{Ierent 

parts ofthe data, and te11s a story. 

NEWYORK CITY'S WEATHER FOR 19AO 

Nd Yd! TiuelJanury 1r,1e&,p.32.

A design with similar strengdrs is Mareyt graphical train schedule 

for Paris to Lyon in rhe r88os. Arrivah aad deparnles from a 

sbtion are locared along the horizontal; length ofstop at a station 

is indicated by the length of the horizontal line. The stations are 

separated in proportion to their actual distance apart. The slope 

of$e line reffeca the speed ofthe trein: the moie nearly vertical 

the line, the faster the nain. The intersection of two lines locates 

the_time and place that trains going in opposite directions pass 

In 1981 a new €xpress nain from Paris to Lyon cut rhe rrip ro 

under chree hours. compared to more than njne hours when Marel 

pubJrshed tfie graphical train scheduJe. The path ofthe modem 

'tc\ (taifl d yafl.le uitesse) is sho*,n, overlaid on the schedule of 

1oo years b€fore: 

GRA?rlIC]lT EXCET,LENCI 31 

E. J. MNy, Ir n.thod. yathique P'n' 

r88r, p.20. Th. ne!}od is nkibured 

to the French eDgineer, !bry.

32 GnA?ItrCAL PR1\CITCE 

The two great inventors of modern graphical designs were J. H. 

L,zllJbert (1728-17j7), ^ Swiss-German scientisr and mathematician, 

and William Playfan (1759-1823), a Scottish political economisi.lo 

The fint known time-series using economic data was published in 

Pfayfairt remarkable book, The Connercial and Political Atlas (Inndon, 

1786). Note the graphical arithmetic, which shows the shifting 

balance of trade by the difference between the import and 

export time-series. Playfair contrasted his new graphical method 

with the tabular presentarion of dxta: 

Information, that is imperfectly acqoired, is generally as imperfectly 

rerained; and a fiun who has carefirlly investigated a 

printed table, finds, when done, that he has only a very faint 

and panial idea ofwhat he has read; aad that like a 6gure 

imprinted on sand, is soon totally eresed and defaced. The 

amount ofmercantile transactio;s in money, and ofprofit or 

los, are capable of being as easily represenied in drawing. as 

any part ofspace, or as the face ofa country; though, till now, 

it has not been anempted. Upon that principle these Charts 

were made; and, while they give a sinple and distinct idea, 

rhey are as near perfrct accurery a< is :ny wey usefuJ. On 

rnspecong eny one ofthese Charts rcentively, a suftciendy 

dntinct impresion will be made, to remain unimpaired for a 

considerable time, and the idea which does remain will be 

simple and complete, at once including the duration and the 

amount. LPages 3 4j 

For Playfair, graphics were preferable to tables because graphics 

showed the shape ofthe data in a comperative persp€ctive. Time- 

IEIXT .'..2 tt. I P2AT'S ad EtPoRI'a b *z t" EnGaAnD 

7..m dz te //a? b n2 4 trlrqlai 

7L tt;tu,t.t t a'4"./,it fDlnt, N. t'br h 4-?'c^t".,4tuont,/?.t1rrt 

w.Lke'..Jry:4d 

- 

t? 

!ol:e TillinC, "E:!ly Expcrihstrl 

Gapht; B.itith Jordal Iot th. Hi'totr 

oIs.itu ,8 l19js), r93-2ra.

series plots did dis, and all but one of the 44 char* in the first 

ediaon of The Connereial atd Politiul Atla: were rime-series. Thet 

one er.ception is the fiIst Lnown bar chart, which playfair invented 

beouse year-ro-1ear dara were missing ard he needed a design ro 

ponray tbe one-year dara rhrr w€rc avarlrble. Nonerhele.s hi was 

skeptical about his innovation: 

This Chan is different Gom the others in principle, as it does 

not comprehend any portion of cine. ard it is much bferior 

inrriLiry ro rhose thrr do: for rhough ir gire, the ex-renr ofrhe 

dill-erent braaches oftrade, it does not cJmpare the same 

branch ofcommerce with irselfat di{fercnt periods: nor do€s 

ir imprinr upon the mbd thar distircr idea,ln doine which. 

dre chiefadvantage ofcLarts consists: for as it wan-e the dimcnsion 

thar ir formed by durarion. rl'ere is no shape given 

to tnc qu.anflnes. lpage 1o1l 

He was right: small, noncomparative, higbly labeled data sets 

usualiy belong in tables. 

tu.r,i,....;?1e dz tu ?..,d,ahe 4.,, tu tL?. ttu 4.*d *.,ttae try* 

The chan does sloq, ar an) rare. Lhe imporn (cross-ha(c|ed 

) a:rd exports (solid lines) to and fiom Scodand in r78r for 

countriet, $.hich are ordered by volume of trade. The horizontal 

is at the top. posibll ro make ir more con!e en( (o see Ln 

ing the points by hand. Zero values are nice\ indieted both 

the absence ofa bar and by a "o." The horizontal scale misrepeats 

"2oo." In nearly all his chars, Pla/air placed the 

ls for the verical scale on the right side oftle page (suggesF 

tbat he plotted the &ta poin* using his left hand). 

Gla?ItIcal EXCEITENCE Jl

34 GRA?IlIC4L ?RACTTC! 

Playfair's last book addresed the question whether the price of 

wheat had increased relative to wages. tn his Iztter on ow agidlltural 

distrcsses, their eauses and renedies; auonpanied uith tables and 

copper-plate tharts shewing and coftFing the lices oI wheat, bread ad 

Iabow,ton 156j to 182r, Playfair wrote: 

You have before you, my Lordr and Gendemen, a chart of 

the prices ofwheat for zjo years, made from official retums; 

on ihe same plate I have traced a line representing, as nearly as 

I can, the wages of good mechanics, such as smiths, masons, 

and carpenters, in order to compare the proportion between 

them and the price ofwheat at every dif,erent period. . . . the 

main fact deserving ofconsideration is, that never at any former 

period was wheat so cheap, in prcportion to mechaaical labour, 

as it is at the present time. . . . [pager u9-3r] 

Here Playfair plotted three parallel time-*eries: prices, wages, and 

the reigns of Bitish kings and queens. 

The hirtory and genslogy ofroyalty was long a gr:phical frvorite. 

This superb construction of E. J. Marey brings together several 

sets of facts about English rulen inco a time-series that conveys a 

sense of the march ofhistory. Marcy (r83o-r9o4) also pioneered 

the development ofgraphical methods in human and animal 

physiology, including studies ofhorses moving ar different paces,

l 

( 

D p6ri( eP, 

GRA?IIICA! XXCtTTINCT 3J 

fie l 

E.I. Mrey, b Mdthode Qdphiqrc @^ni 

1385), 

!. 6, 

I.J. Mdey. Moy.Mr (todon, r89J). 

Segming wnh rhe @cks of tfie horse, 

the timHoies are from pages r91, 224,

36 GRA?HICAL ?RACTICE 

S* 

tl,e movemenr ofa starfrsh turning itselfover (read images fiom 

the bottom upwards), 

the undulations of thc dorsal Iin of a descendinq sea-horse, 

as wcll as the advance ofthe gccko. 

Marey's man in black velvct, photographed in stick-Iigure images, 

became thc time-series fbrerunner of Marcel Duchamp s Nlde 

Delcendiflg a Stihcnse. 

€_ 

ro 

{

The problem with tirne-series is that the simple passage of time 

not a good cxplanatory variable: descriptive chronology is not 

expianation. There arc occasional cxceptions, especially when 

is a clear mech:nism rh:t drives rhe Y-variable. This timedoes 

testiE rbout causality: the outgoing mail ofthe U.S. 

ofRepresentatives peaks every two years,jusr before the 

dav: 

grapbic is worth ar leasr 7oo wo'd:, the number used ir a 

report describing how incumbenr reprerenrariva exploit their 

mailing privileges to advance thei re-election campaiqns: 

TIITE 

YOIINff SflOIIIN 

::{-19iib..tg.9-!!, i! ry! "! !"tq 

-- 

;s;:."ii;j- 

'"LH"l;" 

!rclon h: iis lta 

. ero"d n tr di k !$ 8,. roa 

!e h. bbb, ry o 4:.r b !i ui re 

GRAPTTTCAT EXCETLTNcx 37 

Ftulrlr tu! Er nm .Ef 

rq'q0rr. rw! bF:is-rio ir iodrddd dd j.. ilrJ -hi 5Hr6, 

!r-.ryqY. b- 13 blY --,1.-l!9, 

lron Gh: t+d hd, @iF &d a! id F p6d.4i; 

rls ss

105 r 

t00 I 

35r 

t 05t00..1 

s5.r 

38 GRA?IIICAT ?RAcTICE 

Time-series plots can be moved toward causal explanation by 

smuggling additional variables into the graphic design. For example, 

this decomposition ofeconomic data, a aying r,296 numbel\ 

breaks out the top series into sexonal and trading-day fuctuations 

(which dominate short-terrn chrnged to reveal the long-run trend 

adjusted for inflation. (Note a signi{icant defect in the design, 

however: the vertical grid conceals the height ofthe December 

peaks.) The next step would be to bring in additional variables to 

explain the trarxformed rind improved series at the bo*om.11 

Srtendi. ond nEllb coryone s ofTohr Rdor so,er, untcdSb,es 

I 

----Y 

If tt fl 

l l 

t' 

1 t"f 

r 

111' 11"t' l'11 11 t' TT1'111111'TTl' 

,f 

^..- 

I 

]}"'ffi ,t,,t,,t ,t..t,,t. ,,t.,1.,1, ,t,,t.,t, t,,t.,t, ,t,,t,.t. 

ls tsr l97s l9l 

(" 

f 

V 

/ 

41 

40 

32 

21 

20 

1t05 

ll00 

-95 

I r20 

lt10 

Iro 

rs0 

,.ii"". 

'136 

lrz 

136 

l:r 

t36 

Jn 

!.ii.., 

12 

I 

128 

I ''121 

I 

lzo 

I See Willian s. clscland dd rrma J. 

Teipming, 

"Gralhical 

Methods foi 

Sssonal Adjstn@t," JoMzl d/tlo 

Aberi@ Shritriral ^btiatiaa 77 (M^rdl 

Julius slisLin, "M@suing cur@t E.G 

nomi. IiucturioDs," Sr,rirti.dl -Rd?o/ar 

0uly 1973), p.3.

Finally, a vivid design (with appropriate data) is the before-after 

gb.:a, "l..dt!o,l -?,.D!t; .isnd to.n: av\ q dttruônt" ,a...d b\ a liq^d nt,^s.4 

tu44 .atit,in th. .tu\. fhp Jnap n n.ea.h au, tht"uEh ,r? .upa,r4u, r,p dc?aot 

.? td.cit:okd^. th. ruhp n th" to.D , ,ura.kadt i, iu! t," rt4h, n.Ehnui. b n. d 

rlnd before and after the collapse ofa bridge on the Rh6ne in r84o: 

ponl de Bou.g-SlAndtol sur- le Rh6ne. 

GnAPHTCAT 

!XCrrrtNcE 3g 

M. MitcicU Waldrop, "In Sedch of 

thc Maglciic Monopole,', S.i&.e 

oue a 1e82), p. ro87. 

Charls Joseph Minard, "De ta chure 

des PonB dus 1es grudes Crues,', 

(odober 2a, 1 3J6), Figue 3, in Mimrd, 

coll%tio' de k' bto.h"tet eans,1321- 

186e), held by thc Bibliorhlque dc 

I Ecol€ N,rionale dcaPotus cr Chau$ies.

40 GRAPIIICAL ?AACTICT 

Narrative Graphics of Space and Tihe 

Allr especially efective device for enhancing the explanatory power 

oftime-series displays is to add spatial dimensions to the desigr of 

the graphic, so thet the data are moving over space (in two or three 

&mensions) as well as over time. Thrce e.cellent space-time-story 

graphics illustiate here how multivaiate complexity can be subtly 

integrated into graphical architecture, irtegrated so gently and 

unobtrusivety that;et'ers are \arclly aware that they are looking 

into a world offour o1five dimensions. Occasionally graphics are 

Hligercntly multivariate, advertising the techdqne rather than 

the deta, 3ut not these three. 

The first is the classic of Cbarles loseph Minard (1?8r-i87o), the 

French engiaeer, which shows the tenible fate of Napoleon's army 

in Russia. Described by E. J- Marey as seeming to defi the pen of 

the historian by its brutal eloquence,l2 this conbination ofdata map 

and time-sedes, drawn in 1869, poftIa)'s a sequence ofdevastating losses 

suffered in Napoleont Russian campaign of 1812. Beginning at left 

on the Polish-Russian border near the Niemen tuver, th€ thicl tan 

flow-line shows the size of the Grand Army (4z2,ooo) as it invaded 

Russia in June 1812. The width ofthis band indicates the srze ofthe 

army at each place o11the map. In Septenber, the army reached 

Moscow, which was by then sacked and deserted, with roo,ooo 

men. The path ofNapoleon's retreat froni Moscow is delicted by 

the darker, lower band, which is linked to a temperatue scale and 

dates at the bottom ofthe chart- It was a bitterly cold winter, and 

many froze on the march out ofRussia. As the grapKc shows, the 

crossing ofthe Berezina River was a disaster; and the army finally 

struggled back into Poland with only 1 o,ooo men remaining. AIso 

shown are the movements ofauxiliary troops, as they sought to 

Fotect the rear and the fant ofthe advancing army. Minard's 

graphic tells a rich, coherent story with its multi.variate deta, far 

more enlightening tben just a single number bouncing along over 

time. Sirc variables are plotted: the size ofthe army, its locatiolr 

on a two-dimensional surface, direction ofthe army's movement, 

and temperature on vadous dates during the retreat from Moscow. 

At upper right we se€ Minard's French original, which was printed 

as a two-color lithograph in the form ofa small poster. And at 

lower rigbt, our Englirh translatioD. 

It may well be rbe besr strciscJcal graphic ever drawn. 

12 E.I. Mtey, ra nathnde gaphiqne 

(P!is, 188t, p.71. For more on Minard, 

see AJthu H. RobiNn, "The Thematic 

I'{aps of Cbnl6Jseph Mimrd," Isdso 

Munii, ^ \1967), 95-ro8. 

Utper inage 6on chdl6Joeph MiD'd, 

Tobled* Gophiry\ et Con6 Fig*ati@s L 

M. Mitud, 1845 $6q,Blbhnthiete .le 

Pris, item 28 (62 by 2J cm, or 25 by 10 in). 

E glish t@slation by Dawn Finley ud 

redEving ty llaire Mosq cmpleted

.'--rr/i Jti4aizl:rt r", rN. ",,.,..1." rt.-.,- :. i Q. ,,.. . a,, ..". r.,,.t".:*f,.1. ,- . \,,..,i. rn|r rsr., 

.r..,...,..r,.rri!,.!:.r.r,.-r...!.i....,i.{..,..,r'.,...,..,i.,,,,.,. 

.......,...,,i..-..,....i.......,i.....r.r,,.,-.,,,,....*:..r,"..,.,..,;".......,.,,,.,:.::,.1..t..:1..l-j.:.1:;'" 

,1 

.' 

I 

;tLJf,t ,'tili, iat f ., rri ?;t 

t,!i, a it ttr J N-at$.\t.*^ ,t an 

l:lLllJlL l.tatE "t :j,. nnpabe ?r rirnr o/ r/( Iii.unn. r;?q.d.r];.4,, ri 

I 

: .

42 CRA?IIICAI ?RACTICE 

Thc ncxt timc-space graphic, drawn by a computer, displays 

the levels ofthrcc air pollurants locaed over a twedimensionll 

surface (six countics ir southern Califomia) at four rimes during 

the day. Nitrogen oxides (top row) arc emitted by power plants, 

refneries, and vehicles. Refineries along the coast and Kaiscr Stceli 

Fonuna plant produce dre posr-midnight peaks shown in the first 

pancl; trafic and power plane (with rheir ireary daytime demand) 

send levels up during the day. Crrbon monoxide (second row) is 

1ow after nTidnight except out at the steel plant; morning tra{ic 

rhcn begins to generate each day's ocern ofcarbon monoxide, 

with the greatest concentrarion ar rhe convergence offive fieeways 

in downtown Los Angeles. Reactive hydrocarbons Ghird row), 

like nitrogcn oxides, comc from refineries after midnight and then 

increase with trafic during the day. Each of thc 12 time-spaccpollutant 

siices summarizes pollutanrs for z,4oo spatial locations 

(z,4oo squares five kilometers on a side). Thus z8,8oo pollurant 

readings are shown, except for those masked by peaks. 

Thc air pollution display is a snall n tiple. The same graphicrl 

design structure is repeated for each ofthe twelve slices or rnultiplcs. 

Small multiples are economical: once yiewers understand the 

dcsigl ofonc slice, they have immediate rcces to the dita in all 

the other slices. Thus, as dre eye moves ftom one slicc to the next, 

the constancy ofthe design allows the viewer to focus on chalrgcs 

in the data rather than on changes in graphical design. 

NITBOGEN OXIDES 

CAFBON MONOXIDE 

Las Anqels Tin$, lrly 22, 1979i bêd 

on wor! of cregorr J. Mcxae, Califomia 

La*nuie of TechDologr.

Our third example of a space-time-srory grephic ingeniourly 

mixer space and time on the horizontal axis. This desien moves 

well beyond che conrcndonal rime-'eries becru.e ofiL" clever 

plotting field, with location relative to the eround surface on the 

verric:l axis and rime/spacc on rhe honzonial. Tle life cyde of 

the lapanese beetle is shown. 

More Abstract Designs: Relational GrapLics 

Ile invention ofdata graphics required replacing the laritudebngitude 

coordinates ofthe map wirh more abstrac measures rror 

f,ased on geographicai analogy. Moving from maps to statisticil 

pphics was a big step, and thousands ofyears pased bdore this 

rtep was taken by Lambert, Playfair, and others in the eishteenth 

. Even so, analogics to the physical world served as the 

ceptual basis for early time-series. Piayfair repeatedly compared 

charts to maps and, in the preface to the finr ed.Lt:Lon of The 

nnercial a d Political Atlas, argued that his charrs corresponded 

a physical realization ofthe data: 

Suppose $e money we pay ir any 

_one rcar lo, 'he exp€nce 

of rhe Na!ry u_erc in gunca.. and rhat thesc guirea. wire lajd 

down upon,a lrrgc rable in a strarghr.line._and rouchirg each 

omer. and rnosf paid nexr )ear were latd down in anoiher 

GRAPHTCAI EXCETL!NCr 43 

L. Hugh Neman, Md, a/d hs.h 

(Lordon, 1e6J), p!. ro4-1oJ.

'-d. GRA?IIICAI PRACTICE 

srraieht line, rnd rhe same continued for a number ofyean: 

thesilines would be ofdifi'erent leneths, as there were fewer 

or more guineas; and they would make a shape, the dimensions 

ofwhich would agree exactly with the amount of the sums; 

and the value ofa guinea would be represented by the part of 

space which it covered. The Ch:rts are exactly this upon a 

small scale. and one division represents the breadth ot value of 

ten thousand or an hundred thousand guineas as marked, with 

the same exachess that a square inch upon a map may represent 

a square mile ofcountry. And they, therefore, arc a representation 

of the real money laid down in diFerent lines, as it was 

originally paid eway. lpages iii-ivl 

Fifteen years later in The Statistual Breviary,his most theoretical 

book about graphics, Playfeir broke free ofanalogies to the physical 

world and drew graphics as designs-in-themselves. 

One of four plates in The Statisdral Bteviary, rhis graphic is distingunhed 

by its multivariate data, the use ofarea to depict quantity, 

and the pie chart-in apparenrly the 6nt application of these 

devices. The circle represenrs dre area ofeach country; the line on 

the left, the population in rnillions read on the vertical scales; the 

line on the right, the revenue (taxes) collected in millions ofpounds 

sterling read also on the vertical scale; and the "doaed lines drawn 

between che population and revenue, are merely intended to connecr 

together the lines belonging to the same country. The ascent 

SIATISTICAI BREVIARY: 

B'FrLLl'aPL'fFlIL 

s-^/: /rr,"/,* npth " ".?-.,,--V* "-lt:*t 4;',",r.-"-.",' ,.,., 

6 

xDx

ofthose Iines being ftom dght to left, or from left ro right, shews 

whether in Foportion to its population the country is burdened 

with heary ta-res or otherwise" (pages r3-r4). The slope ofthe 

dotted line is unbformative, since it is dependent on the diameter 

ofthe circle as wellas the heighr ofthe two v€rticals. Hov/evet, 

the sign ofthe slope does maLe sense. taking Plafan to his familiar 

point about whit he rcgarded as excessive taxation in Bdtain 

(sirtb circle from the ngbr. wich rhe dope runlrrjng opposire Lo 

mosr countriet. Playfair was enthusiastic about the multivariate 

arrangfment becaLise ir fos(ered comparisom: 

The autlor ofthis work applied the use of Lines to matters of 

commcrce and Erance abour filteen lears ago. -rrh grc,r 

succels. HJs mode wa" qenerallv approved ofas nor o'iiu fr.itir.rting, 

bur rcnderirg rlose srudie' rno'e cle,r, and rerjrned 

more e.srl) b) Lhe memo1. The pre'enr charL. rrc in like 

mamer inrerded ro aid rror'rjcal studies. bv shewine ro $e 

eye Lle 'izes ofdifferent courrrier rcpre.enrid bf imll,r form'. 

for where forms are not similar, the cye cannot iompare them 

easily nor accurately. From this circuirxtance it hapliens, thar 

we have a more accunte idea ofthe siza ofthe olaie*. whi.h 

are rphe,es. rlran of rhe narion. of Europe rn hich ". '* ." 

thc maps, all ofwhich are irregular forms in themselves as well 

as unlrke to each orher. Sizc. Populanon, and Revenue, are rhe 

rluee pnnopal objecrs ofarrenrion upon rbe gcncral scale of 

't.rti.rical srudie., wherhe, wc are acrurred bv"curiosirv or 

inrerev: I h:r c rherelore represcnred these cbrec objeci' in one 

r'lev/.... L?age 1Jl 

But here Playfair had a forerunner-and one who tJrought more 

clearly about the abstract probicms ofgraphical dcsign than did 

Playfair, who lacked mathematical skills. A most rem:rLable and 

explicit early theoretical statement advencing the gereral (nonanalogical) 

relational graphic was made byJ. H. I-ambert in 1765, 

3J y€ars before The Stutistical Breuiaryl 

We have in general two variable quantities, x, y, which will 

be collated with one another by observation, so rhat we can 

determine for each value ofx, which may be considered as an 

abscissa. rhe correspooding ordinare y. Were rhe expenments 

or observat;on\ comDletelv l(currle. the.e ordinateicould 

qire a number ofpoinrs throush *hich a straishr or curved 

Ine should be dr";n. B.,t ". dis s not 'o. theine deviarer ro 

a grcaler or lesse! exrent from the observational poinrs. ft must 

thercfore be drawn in such a way that it comes as near as 

possible ro is rrue posirJon dd goes. as ir u ere. tluough thc 

mrcldre ot the elven Dol]lts.', 

cRA?EaCAt TXCtLTENCE 4J 

!3johm Hetuich Lamhert, Bdlr,ile 

zen Cebtuucla da Mdthtuotik u"d d,th 

,r/dd!r8 @erlin, 1?6t, s quored iD 

Lam Til1iig. Earry nxlcjfteqral 

Gapb.," Btxbh Joanal ht the Histotr 

oJ S.iena,I (1s7l), 2o4-hs.

46 GRApETcAL ?RAcTrcr 

Lambert drew a graphical derivation ofthe evaporarion mte of 

water as a function oftemperature, according to Tilling. The 

aralysis begim with two rime-seies: DrF, showing the decreasing 

height ofwater in a capillary tube as a fuaction of time, and alc, 

the temperature. The slope ofthe curve DrF is then aken (note the 

targent Dr c) at a number ofplaces, yielding the rate ofevaporation: 

To compiete the graphical calculus, the measured rate is plotted 

against the corresponding ternperarure in this relational graphic: 

Thus, by the early r8oos, graphical design was at last no longer 

dependent on direct analogy to the physical world-thanLs to the 

work ofl,ambert and Playfair. This meant, quite simply but quite 

profoundly, that any variable quantity could be placed in relationship 

to any other variable quantity, measured for the same 

units ofobseffation. Data graphics, because rhey were relational 

and not tied to geographic or time coordinates, became relevant 

J. H. &mbs! 

"Essi 

d'hygrom6rrie ou 

su li me@e de I'hlmidi4" Mlrotar 

de I'Aat#6ie Royle d?s S.ie&s d Bdl.r 

Lenes . . . i769 @*li^,177r), plare t 

facins !. 126; fron Tilingt artide.

to all quantitative inquiry. Indeed, in mod€rn scientific literature, 

about 4o percent ofpublished graphics have a reiational form, 

with two or more variables (none ofwhich are latitude, longitude, 

or time). This is no accidenr, since the relational eraphic in its 

barev lorm. (ne \crrrerplor and i15 \ari:nts is rle gieacesr of:ll 

graphical designs. It iinls at least rwo variablcs, ercouraging and 

even imploring the viewer to assess the possible causal relationship 

betwecn the plotted variables. [t confrons causal theories that X 

causes Y with empirical evidence as to the acrual relationslLip between 

X and Y, as in the case ofthe relationship berween hns 

cencer and smoking: 

CRUDE MALE DEATH RATE FOR LUNG CANCER 

IN T95O AND PER CAPITA CONSUMPTION OF 

CIGARETTES IN 1930 IN VARIOUS COUNTRIES, 

DEN' 

/j 

CELAND 

alFsweorr I 

./, 

ND 

7 

CI6ARETTE 

CONSUMPT]ON 

zt7,t2'a)72:r',4-.at:z,atzt,,/-- 

Relorr of ihe Advis.ry Commirta to 

the Surgeon Generrl, S-oliry drd tledlr,t 

(War\iDgton, D.C., 1964), p. 176; based 

on R, Doll, "Etiotogy ofl-ug Cdcer," 

Adudne' if Cdaat Raurh, 3 (rsi),

48 GRTTPEICAL ?RACTICB 

Thesc small-mu1tiple relitional graphs show unemplolanent and 

inflarion over time in "Phillips curve" plo* for nine countries, 

demonstrating the collapse ofwhat was once thought to be an 

inverse relationship between the variables. 

ln8doo rd UMDlotnest Rds 

Unit€ds€t.3 

Patl, Mcct .ke, .t d., Towads FUI 

Enplornen ann ht@ Stabiw lPais,

Theory and measured observations diverge in the physical scialso. 

Here the relationship berween temperature and the 

conductivity ofcopper is asessed in a series ofmeasurefrom 

diferent laboratories. The connected points are from 

single publication, cited by an identfication number. The very 

answers reponed in the published literature result mainly 

impffities in the semples ofcopper. Note how effectively 

graphic organizes a vast amounr of alata, recording findings 

Lundreds ofstudies on a single page and, at the same time, 

ing comparisons of the varying results. 

GRAPlrrCAL rXCrr!xNCE 4g 

C. Y.Ho, R- \v. Poqell, mdP.n.riby, 

Thmal Condaxtiviry oJ th! Elenuk: A 

Coryehîue Reritu, snppletunt to. 

r, Joauat of Pfusnat ann ctukkdl 

Ref,reNe Data, 3 (914,1 2aa.

JO GRAPEICAL PRACTICI 

Finally, two rclational designs ofa diferent sort-.rherein the 

data points are thernselves &ta. Here the e{fect oftwo variables 

intencting is portrayed by the fues on the plotting ield: 

And similarll tle varying sizes ofwhite pine seedlings after 

growing for one seson in sand containing di1rerent amounts of 

calcium, in paro per million in nutri;ntland ctltures:.. 

E- C. Z@a!" 

"Cat 

itlophe Th@ry." 

S.i6tijc Z6eti.M, 2a1" ( prll 1976), 67i 

baFd @ K(md Z. Lor@, Kir{ 

Sot M'rairy Nry Yqk, 19Jr). 

H. L. Mitddl, me Gtovth ann Ntfiinn 

oJ Whit Pire Seedling if Crlxtcs wih 

Vdthg Nittog.r" WtI'hM, Potasiut 

dd CdLiM, The Black Rock ror6t 

Suleti{ No. 9 (Cm*zll{nih+ 

HuiL@, NN Yck, 1939), ?. zo.

Principles of Graphical Excellence 

Graphjcal cxcetlencc ;s the \\,c1l-designcd prescmtarjon of intcrcsring 

data-a maaer of rrrLrrarrrc, of srnrisrirs, and of r/r-.iqrr. 

Graphical crcellcnce consnts of comptex tdcas communlcatcd 

rith clariry, prccision, and eliiciency. 

Graphical cxcellence is that nhich givcs to thc viewcr fie grcatesr 

number ofideas in the shortesr limc widl thc least nrk rn thc 

smallcst space. 

tt 

rT-r 

Graphical cxcellence is nearl,v alu,ays rnultivariate. 

And gr:rphical 

excellence 

requircs 

telhng dre truth about the data. 

cIAPHJCA L !XCTLLINCt J1

As to the roptiery and justness of rcprcsenting suns oI noneL and tine, 

by pats of spaee, tho' very rcadily agreed to by no* net, yet a Jew seen 

to q?rchend that therc hlay possibly be sone deception in it, of which 

they are not awdle. , , . 

Willian Playfair, Tlz Coanerial dnd Palitial AtlB (Lonaan, 

t86) 

People said: " lvith the chd/t ot1 the uall, uith the j&utes pullished, let's 

emdate and nuse otr etthusiasn in prodution." 

St*e St.tistic.l Bureau ofthe Peo?lei Relublic of Chitu, 

Statitt 6l Wotk in the N.u China (Beijing, 1979) 

Get it right or let it alone. 

The coftlusion fou 

junp to nay be your own. 

Jdnd Thurber. F,rrra FdrLrrl o,' Ti,s (Ncw York. 1e56)

2 Graphical Integrity 

For many people the frrst word that comes to mind when they think 

about statistical char* is "lie." No doubt some graphic do distott 

the underlying data, making it hard for the viewer to leam the 

truth. But data graphics are no different from words in this regard, 

for any means ofcommunication can be used to deceive. There 

is no reason to believe that graphics are especially vulnerable to 

exploitation by lian; in fact, most ofus have pretty good graphical 

lie detectors that help us see right through frauds. 

Much of twentieth-century thinLing about statistical graphics has 

been preoccupied with the quesrion ofhow some amateurish chart 

might fool a naive viewer. Other important issues, such as the 

use of graphics for serious data analysis, were hrgely ignored. 

At the core of the preoccuparion $drh deceptive graphics was the 

a$umprion that data gnphics were mainly devices for showing 

the obvious to rhe ignorant. It is hard to imagine any doctrine 

more likely to stifle inrellectual progres in a field. The assumption 

led down two fruidess paths in the graphically barren ye:n 

from r93o to r97o: First, 

"alive," 

thrt graphics had 

"com- 

to be 

municatively dynamic," overdecorated and cxaggerated (otherwise 

all the dullards in the audience would fall rsleep in the face 

of rhose boriDg statistics). Second, rhat the main task ofgraphical 

analysis was to detect and denounce deception (the duliards could 

not protect themselves). 

Then, in the lrte r96os, John Tukey made statistical graphics 

respccbble, putting an end to the view that graphics were only 

for decorating a few numbers. For here was a world-class dara 

analyst spinning of half a dozcn new designs and, more importantly, 

using them effective]y to e]?lore complex data.1 Not a 

word abour deception; no tortured anempts to consEuct more 

"graphical 

standards" in a hopeless efort to end all distorions. 

Instead, graphics were used as instrumen* for reasoning about 

quantiredve information. With this good example, graphical work 

has come to flourish. 

Ofcourse false graphics are still with us. Deception musr always 

be confronted and demolished, even iflie detecrion is no lonqer 

ar rhe forefront ofresearch. Craphrcal excellence beg'ns wirh;lling 

rhe truth ebout the data. 

ljohn \v. Tutey rd Martin B. Wilk, 

"Dda 

Anllylis &d Sbristica: Tcchniqu6 

dd Appro.ch6," id Edward R. Tufte, 

ed., The Qudtit^tit. Anald' oJ Srict 

&otiem Ee.ding, M:s., r9?o), 3?o- 

39o; ud]ohr'w. Tukey, "Some 

Gr2phic &d S.migmphic Displ.ys." 

-l 

in 

. L. Bnda{t, .d., sldtndut P4?4s ln 

Honot oJc.otee w. stud{ot ( m6,

J4 GRAprrrCAl !RACTTCI 

Hcre are several gmphics that feil to tcll the truth First, the case 

of the dnrppcarnrg baselinc in the annual report ofa company that 

wouldjust as soon forget about r97o. A careful look ar the middle 

panel reveals a ncgative incomc in rgzo, which is disguised by 

having the bars begin at the bottom at apprcximately minus 

$4,2c1o,ooo: 

This pseudo-dccline was created by comparing six months' 

worth ofpaymcnts in 1978 to a full year's worrh in 1c76 and 1977' 

with the 1ie repeated four timcs over. 

D^y Mn\es, L1c., t97 Ann dl Repn,P.1, 

N.u Yrl T/n6, Augrst 8, 1978, 

!. c1.

And sometimes thc fact drat numbers have a magnitude as well 

as rr order is simply forgotten: 

compa€ttve Annurl (;6t pd c.pib fd @ of tnM. in 

Pftrb!4h crty Hom6 ud Penn yl*trt, stare H6pihh, 

,r47 arn $93 

\what is Distortion in a Data Graphic? 

A graphic does nor distort if .he visual representation ofrhe data 

is consistent with the numcrical representation. \Vhat then is the 

"visual 

represenration" ofthe data? As pLysically measured on the 

surface of the gnphic? Or the pereiled visual e6ecr? How do we 

know that the visual image represents rhe underlying numbers? 

One way to try ro answer these quesrions is to conduct cxperiments 

on the visual perception of graphics - Laving people look 

at lines ofvarying length, circles of different areas, and then 

rccording rheir a*e-mcnt. of rhe numeric:l qLranririe.. 

I tlink I 3e tbat d@ B 

ir 3.ia tim6 biggq thd 

dq A. L rhat txfiec1? 

Such cxperiments have discovered vcry approximate power laws 

relating the numerical measure ro rhe reporred perceived measure. 

For exrmple, the perceived arca ofa circle probably grows somewhat 

morc slowly than the actual (physical, measured) arca: 

the reported perceived area : (actual area)*, where x : .g+.3, 

a discouraging result. Dilicrent people see the same areas somewhat 

CRAPIIICALINTTCRITY JJ 

Piilsburgh Civic Comni$ion, Re?or o, 

Expentlihr.' o.f the D?pathht rJchdtitiet 

(Pittsburgh, 191r), t. 7.

JO CRAIIIICAt PtACTTCt 

differently; perceptions change with experience; and perceptions 

are context-dependent.' Particuiarly disheartening is the securely 

established finding that the reported perception of something as 

clear and simple as line lengrh depends on the context and what 

other people have alrcady said about the 1ina.3 

Misperception and miscommunication are certainly not special 

but what is a poor designet to do? A different graphic for each 

perceiver in each context? Or designs that correct for the visual 

trensformations ofthe average perceiver participating in ihe average 

psychological experiment? 

One satisfactory answer to these questions is ro use a table to show 

the numben. Tables usually outperforn graphcs in reporting on 

small data sets ofzo numben or less. The special power ofgraphics 

comes in the display oflarge &ta sets. 

At any rate, given the perceptual dificulties, the best wc can 

hope for is some uniformiry in graphics (ifnot in the perceivers) 

and some assurance that perceivers have a fair chance ofgening 

the numben right. Two principles lead toward these goals and, in 

consequence, enbance graphical integrity: 

The representation ofnumbers, as physically 

mea$red on the surfacc ofthe gnphic itsell 

should be directly proportional to the numerical 

quantities rcpresented. 

Clear, detailed, and thorough labeling should be 

used co defeat graphical distortion and ambiguity. 

rvrite out explanetions ofthe data on the 

graphic itself. Labcl imporhnt eventi in the data. 

,Thc drmsivc litemtue is summdized 

in Michael Macdonald-Ros, How 

NuDters Are Shom: A ReYiew of 

R.smrch on the Plefnaiotr of Qlnrit 

tive D.b in Tds," /c/iGt.val 

Conûnndtian Rerieu, 25 i.1977), 359 

4@. In larticular, H. J. Meihoefe. fnds 

sr€at vriabilitr among p.(eivasi rc 

"The Utility ofthe Clrde as d E6ecnvc 

Canogrlphic Symbol," C6ddtar Crtott'phd,6 

11969), 1as-1lji nd "'lhe 

Visul Pe.ception of the Cnde in The 

matic Ma!s: Expdihfttel Re$lts, 

ibid., 1o (1971), 63-84. 

3S. E. Ash, "Studi6 of Irdepdddcc 

rnd Submision to Group PresE. A 

Minority of One Against a Unmimou 

M^jotlty. pr.hologi&l Monostarhs 

DFwing !y CEM; colt'right 1961,

Violations ofthc first principlc constitute one form of graphic 

misreprcsentation, measured by the 

Lie Fa*or : 

size ofeffect shown in graphic 

size of effect in data 

If the Lie Factor is cqual to onc, rhen the graphic rnight be doing 

a reasonablejob ofaccurately representing the underlying numbers. 

Lie Factors gr$ter than 1.oJ or less rhan .9J indicatc suL 

stantiai distortion, far beyond minor inaccuracies in piotting. 

The logarithm ofthc Lie Facor can be taken in order ro compare 

overstating (iog LF > 0) with understating (log LF < 0) crrors. In 

practice almost all distorrions involvc overstaring, and Lie Feoors 

of two to Iive are nor uncommon. 

Here is an extreme example. A newsprper reported that the 

U.S. Congres and the Deparrment ofTransportation had set a 

series offuel economy standards to be mct by automobile manufacturers, 

beginning wirh r8 rniles per gallon in 1978 and moving 

in steps up to 27.j by 198J, an increase of J3 percenr: 

275-li!^ 

roo - s.rc" 

lg.0 

These strndards and the dates for their anainment were shown: 

This linc, rep.senxng 18 mil6 per 

8.Ion in 1978, is 0.6 inches long. 

This line, rep.senting ,7.J niles pa 

g.Uon in 198r, is J.l incles long. 

GRAPHICAT INTTGRITY J7 

Frcl Eonomy Standards 

lor Autos 

Sel by Cdrgrcs eid slDlrendted by ho TnnrDortario. 

o€larlno.l. h hiler rer g.llon. 

N., Yol& It'6, Augut 9, 1978. ?. 12.

J8 cRAPIIIcAL ?RAcTIcE 

The magnitude ofthe changc from 1978 to 198J is shown in the 

graph by thc rclative lcng.hs of thc two lires: 

51-n6 ^ lrn : 7819" 

Ll.l) 

Thus the numerical change of 53 percent is presented by some 

lines that changed 783 percent, yielding 

which is too big. 

LieFactor:E: r+.s 

53 

lJre dr,play J'o ha, 'everoJ peculrrnre' ofperrpecrire: 

' On most roads the future is in front ofus, toward the horizon, 

and the present is at our feet. This display revenes the convention 

so as ro enggerate the severity ofthe rniieage strndards. 

. Oddly cnough, the dates on the left remain a constant size on 

thc page cven as they rnove along widr the road toward the 

horizon. 

. The numbers on the right, as well as the width ofthc road 

itself, are shrinking because oftwo sinultancous effects: the 

change in the values portmved and the changc due to penpective. 

Viewers havc no cha:rce ofseparating the two. 

It is easy enough to dccoratc these data without lying: 

UAUIRXD FUtrL ICONOMY ST^NDATDS: 

NEw c^Rs BUrr rRoM 1978 ro 198J 

19?8 1979 1980 1981 1982 i98l 1984 198J

The non-lying version, in addicion, pu* the data in a contcxt by 

comparing the new car standards with the mileage achicved by 

the mix ofcars actually on the road. Also revealed n a side of the 

data disguised and mispresented in the original display: the fuel 

economy standards require gradual improvement at star-up, 

followed by a doubled rate from 1980 to 1983, and llatteting out 

after thrt. 

Sometimes decoration can help ediorializc abour rhe substance 

ofthe graphic. But it is wrong ro disrort rhe dara measures rhe 

ink locating values ofnumbcn-in order ro make an cdirorial 

comment or lit a decorative schemc. It is also a sure sign ofthc 

Graphical HacL at work. Hcre are many decorations bur no lies: 

CRA?IIIC4LINTTGRITY J9

OO GRAPIIICAL PRACTICE 

Design and Data Variation 

Each part ofa grrphic generates visual expecotions about its other 

parts and, in the economy of graphical percep tion, these expeciations 

often determine what the eye sees. Deception results from 

the incorrect extrapolation ofvisual expectations generated at onc 

place on the graphic to othcr placcs. 

A scalc moving in rcgular intcrvals, for cxample, is expccted 

to continue its march to the vcry end in a consistent faslion, without 

the muddling or trickery ofnon-uniform chrnges. Here an 

irregular scale is used to concoct a pseudo-declinc. The first sevcn 

increments on the horizontal scale are ten years long, masking 

the rightmost interval offour years. Consequendy the conspicuous 

feature ofthe graphic is the apparent fall ofcurvcs at thc right, 

particularly thc dcciinc in prizcs won by pcople Gom thc Unitcd 

States (thc hcavy, dark linc) in thc most rcccnt period. This cffcct 

results solely Gom dcsign variation. lt is a big lie, since in reality 

(rnd even in exrrapolation, scrling up each end-point by 2.J to 

take the four years worth ofdata up to a comparable decade), 

the U.S. curve turned sharply upward in the post-197o interval. 

A correction, with thc actual data for r97r-8o, is at the right: 

l{obel Prizos Awarded in Science, 

lor Seloclod Countrics, 1901-1974 

1921- 1931 r94r- 1951- 

1920 1930 1940 1950 1960 

30 

Nariomi Scida loudaiion, S.me 

.rd@to6, r97, Cwashinston, D.C., 

l{obel Prizes Awarded in Science, 

lor Selected Counlries,lg0l-1980 

l90r l9]]- 1921. 1931 1941. 1951. 1961 197r. 

1910 1920 1930 1940 1950 1960 1970 1930

Thc confounding of lesigfl 1)anado with data vaidtion o'er the 

surface of a graphic leads to ambiguity and deception, for rhc eye 

may mix up changes in the design wnh changes in the data. A 

steady canvas makes for a clearer picture. The principle is, thenr 

Show dara variation, nor design vanaoon. 

Design variation co upts thn display: 

OPEC Oilhlces: Atter 18 

tlonth3 olslability, Pric$ Are *[;{ 

DuetoftseAlnin -,.,r; 

Ooll.nprb.tr l 

Five diferent vertical scales show the price: 

During this time 

19?3 r9J8 

January March 1979 

Ap l June 1979 

Juiy September 1979 

October December 1979 

one vertical inch equels 

$8.oo 

$4.73 

$4.37 

$4.rr 

$3.92 

And two dilferent horizonkl scales show the passaqe of rimc: 

During this time 

1971-197E 

1979 

one horizontal inch equals 

3.8 years 

o.J7 years 

As the two scales shift simultaneously, rhe distorrion rakes on 

multiplicative force. On the left ofthe gmph, a price of$lo for 

one year is represented by o.3r square inches; on rhe righr side, 

by 4.69 square inches. Thus exacdy the same quanrity is 4.69/0.31 

:1i.1 times larger depending upon where it happens to fall on 

the surface ofthe graphic. Tlar is design variarron. 

*s 

Quarterly 

CRAPIIICALINTECRITY 61 

N./ Yo* Tffi.r, De.ehber 19, 1978,

62 cRA?HIcAl PRAcrrc! 

Dcsign variation infected similar graphics in other publications. 

Hcrc an increase of+s+ percent is depictcd as an increase of4,28o 

perccnt, for i Lie Factor of9.4: 

And an increase of7o8 percent is shown as 6,700 percent, for a 

Lic Factor of9.j: 

55i:if:5*.?f: til'-1;g;,trr,** OPEC Be clnr|:llt Pris 

=:-.5.=:.3;A:; :q:*-."":.:.::ii 197019;9 

*;:.eJ,'* '- --=,*r'-:.'.}!.f ' "::j.- aF 

Ali these rccounts ofoil priccs made a second error, by showing 

the price ofoil in inflated (currcnt) dollars. Thc 1972 dollar was 

worth much norc then the 1979 doilar. Thus in sweeping from 

Tine, Lptil9, 1979, p. s1. 

Wthirytan Po!t, M*ch 2a, rs7g, p.

left to right over the surface ofthe graphic, the vertical scale in 

efcct changcs-design variation-because the value of money 

changes over the years shown. The onlv way to think clearly about 

moneyovcr hme is to make comparisons udng inflation_adjused 

units ofmoney. Several distinguished graphc designers did ex_ 

press the prjcc in real dollars-and they also avoided other sources 

ofdcsign variation. The stars were Brrinc$ Iteeh, the Sunday 

Tinrr (London), and The Econanist. 

700 

500 

500 

.100 

100 

:00 

pnce of crude 

oil 

I 

:100 | 

t\0[1lNAL 

In the graphic wc saw first, the rwo sources ofdesign variation 

covered up rn intriguing, non-obvrou. a"pecr ofrhi d-ar:: irr rhe 

rour years Frtor ro the 1979-1980 increases, rhe rcal pricc of oil had 

lecllnel. Busy with decoration, tLe graphic had mised the news. 

kddsdRy'fuh{],i 

hroft.&- . _. .,j- 

_- 

GRAPI{ICALiNTTCRITY 63 

s0ft hrufi 

Th. tutunist,D.cc1ltr 29, 19:/s, p. 41. 

Sdr'dr Tir.r (London), Dccenber 16, 

Brsitr s Week, Aprjl s, rs7g, p. g.

71e Dislonr at the-Bolram are Ear+ kthar on the ]tuqLt hand Mdl)''

TLe Case ofSkyrocketing coveftment Spending 

Probably the most frequendy printed graphic, other rhan rhe daily 

weather map and stock-market trend line, is the display ofgovemment 

spending and debt over the years. These arrays nearly 

always create the impression that spending and debt are r:pidly 

increasing. 

As usual, Playfair was the first, publishing this finely designed 

gnphic ir 1786. Accompanied by his polemic againsr the "ruinous 

folly" of the British govemment policy of financing is colonial 

wan through debr, it is surely rhe 6rst skyrockering govemment 

debt chart, beginning the now zoo-year history ofsuch displays. 

This is one of the few Playfairs that is taller then wide; less thaa 

one-tenth ofail his graphics (about 9o, drawn during 35 yeats of 

worl) are longer on the vertical. The tall shepe here serves to 

€mphasize the picture ofrapid growth. The money figures are 

not adjusted for inflation. 

But Playfair had the integriry to show an altemative venion a 

few pages later in Tlre Coulttlercial and Poltical Atl6. The inrerest on 

the national debt was plotted on a broad horizontal scale, diminishing 

the skyrocket effect. And, furthermore, "This is in real and 

not in nomiml millions" (page rz9): 

]ftd of dD NATIONAL DEBT f!!D .h€ R.Nl[tid. 

Tttz totun bu 6 r@t.ra* a4 bt,qib h@r'.r'tt w 9a "dazr. 

GRA?IIICA! INT!GRITY 6J

66 cRA?IIrcAr PRAcTrcr 

Although Playfair deflated money units over time in his work of 

1786, the matter has proved difrcult for many, eluding even modern 

scholars. This display helps its political point along by failing 

to discount for infatron and population growth and by r:sing a 

tall and thin shape (the area covered by the data is 2.7 times taller 

tharr wide): 

Let us look, in detail, at another graphic orr government spending: 

New lbd( State 

Toral Budget Expenditurcs and 

Aid to Localities 

Total Btrdger ---t 

t t 

Mafti. Fiot'tu, Congst: Kelttov of nE 

wthinStB BrablishMt G:tew Hâa\ 

1917). p.92. 

Ntu Yotk Tin*,Feht@ry \ 9j6, p.

Despite the appearance created by the hlaeractive design, the state 

budget actually did not increase duing the last nine yean shown. 

To generate the thoroughly false impresion ofa substantial and 

continuous increase in spending, the chart deploys several visual 

and statistical tricks-all working in the same direction, to exaq_ 

gerate rhe growrh ir rhe budger. These graphical gimtlicks: 

This dusrer of rylc ehphasias &d 

sEerchcs our the low hlu for 1966- 

1967, a.ouaging thc impression tbai 

redt yem l,ve sttor u! ftom a shall, 

statle base. Horizontal do*s ?rovide 

Totat Budser --r P 

TorltAid to 

Lo.atitiss. 

Tlis squeerd-doM blo& of ryF 

contributes to an imse of small, 

sq@ered-dom rufuers b,ck in the 

Leaving behind the distortion in the charqjunk heap ax the lefr 

yields a calmer view: 

/|lill 

,rtlllillll 

CRA?IIICAL INTEGRITA 67 

These tlJe par.lelepipcds }ave te@ 

!la@d otr u olriei llme irto,t 

of the othd eight, dmring the imrge 

rlat the Dewer budgets towe. over rhe 

r 4 t 1 s 1 6 n 

i t 

Anows poinring sEaisht up emptasize 

r.ff, g'oMb. ComlaR wirh honzonut 

.row at left. 

ilrrffillll

68 cR,{?IIrcAL ?RAcTIcE 

Two statistical lapses also bias the chart. First, during the years 

shown, the state's population increaed by r.7 million people, or 

10 percent. Pert of the budget growth simply paralleled population 

growth. Second, the period was a trme ofsubstantial irflation; 

those goo& and services that cost state and local govemmcnts 

gr.oo to purchase in 1967 cosr $2.q in 1977. By not deflating, dre 

graphic rrixes up changes in dre value ofmoney with changes in 

tJre budget. 

Application of aithmetic makes it possible to take population 

and inflation into account. Computing expenditures tn constant 

Qeal) dollars yr capita tereals a quite different-and far more 

$360 - 

$34o - 

budg€i extdditurs, 

in .6tant dols 

1967 1968 1969 r97o 1971 1972 1973 19% 19jS 1y76 r97? 

Thus, in terms of real spending per capita, the state budget increased 

by about zo percent frorl.r 7967 to 7g7o and remained 

relatively corutant frorn r97o drough 1976. And the 1977 budget 

represents a substential ledine in expenditures. That is the real news 

story of these data, and it was conpletely missed by the craph of 

the Magical Parallelepipeds. Ofcourse no small set ofnumbers is 

going to capture the complexities ofa large budget-but, at any 

rate, why tell lies? 

The prbciple: 

In time-series displays ofmoney, deflated and 

standardized units of monetary measnremetrt ale 

nearly always better than nominal units.

Visual Area and Nurrerical Measqre 

Another way to confuse data variation with design variation is ro use 

areas to show one-dimcnsional abta: 

Acuoissemenl de nos 

etporlations d'aulos 

1927 - 1929 

And here is the incredibie shrinking doctor, with a Lie Factor of 

2.8, not counting the additional exaggeration from the ovcrlaid 

penpective and the incorrec horizontxl spacing ofthe data: 

TIIE SHRII{KIIIG FAIIII[Y DllCT(IR 

h Crllfornlr 

Psrcenhge 0l lhctors hvotsd Sohty io Famity practic€ 

1:2,247RnnoDpopu,lro[ 

8,023 DGlds 

GRA?IIlCAI INTTCRITY E) 

R. Satct, Iir G/a?t'iq,er eris, 1932). 

Iat Afg.le! Tin6, Angrsr s, 1979, p. 3.

70 GRA?HtC41 ?RACTTCI 

Many published e{Iorts using areas to show magnitudes make 

thc clcmentary nistake ofvarying both dimensions simulraneously 

in rcsponsc to changes in one-dimensional data. Typical is dre 

shrinking dollar fallacy. To depicr rhe rare ofinllation, graphs 

show currency shrinking on rwo dimensions, even though the 

value ofmoney is one-dimensioual. Here is one ofhundreds of 

TIt tt (irtrtqr), strlrslt ll,\r|'xrr!tr.! 

-'ffi 

TI lr l ,\r{cr ) !it,,!Tt:s o!.,\!}xu( -\ 

rrlt (ft!E9{l!!E! rlt .!rLP.r!!J 

.r_ 

"@ffiiil 

ilrj-:'IWWL-.*? 

'o t.s i 

Ifthe area ofthe dollar is accurately to reflect its purchasing power, 

then the 1978 dollar should be about twice as big as that shown. 

W$hitStan Pa{, O.tohet 2t, 1978. P. !.

There are corxiderable ambiguities in how people perceive a twodimeruional 

surfaceand then converr rhar percepoon inro a onedimensional 

number. Clanges in physical area on rhe surface ofa 

graphic do not reliably produce appropriately proponional changes 

in perceivcd areas. The problem is all the worse when the arexs 

are tricked up into rhree dimensions: 

By surface area, the Lie Factor for this graphic is 9.4. Bur, ifone 

takes the barrel netaphor seritrusly and assumes that 6e uolune o{ 

the barrels represents the price change, then the volume ftom 1973 

to 1979 ncteases 27,ooo percent compated to a data inqease of 

4J4 percent, for a Lie Factor of 59.4, which is a record. 

Similarly, a tltee-dimensional represenration pu6ng up 

oDe-{lmensloIr3t data: 

Conclusion: The use of wo (or three) varying dimensions to 

show one-dimensional data is a weaL ald ineficient rechdque, 

capable ofhandling only very small &ta sets, often v/irh error in 

desiga and ambiguity in perceprion. These d€signs cause so many 

problems drat they should be avoided: 

The number of information-carrying (variable) 

dimensions depicted should not exceed the 

number ofdimensions in the &ta. 

GRAIHICAT INTTGRITY 71 

Ntu Yotk Ti66,Jetty 2j.1981,

I 

t , 

CASSEPOSTALI 

DI RISPARMIO ITALIANE 

Numerodei 

Lrbretti, Libretto medro e Deposlto totale 

al frne di oqni mese 

t Err Ei9R L 

This multivariate history ofthe Italiaa post ofrce uses two dimcnsions 

in a way nearly consistenr with this principle, with the 

number ofposal savings books issued and the average size of 

deposia multiplying up to total deposits at the end ofeach montL 

froE 1876 ro 1881. 

l@.otuo G.bag\o, T.6ia CtuaL delle 

Sr.rirri.a (Mil2n. se.ord editior, 1888).

Bur Plavlair's circles, an earlv usc otarea to shoiv mrenirudc, 

arc nor consistcnr rvidr rhe pnnciplc. snrce thc txc-dimensioral 

data (ciry popularions) are reprcscnted bl an arcal dara ncasurc: 

Perhaps graphics that border on cartoons should bc crcmpt 

liorn thc prnrciplc. Wc certainlv would not rvant to forgo the 

.1,34o pound chickcn: 

I 

J 1,4 

i! 

tl 

i 

t 

cnA PHlc4L I N T!CRtT Y 7l 

sti4 if. A,rtuitt R?J,a,a Baok lNe\| 

t.

125 

74 GRAPHICA! !RACTICE 

context is Essential for Graphical Integrity 

To be truthful and revealing, data graphics must bear on the quesrion 

at the heat ofquantitarive thinking: "Compared to what?" 

TLe emaciated, data-thin design should always provoke suspicion, 

for graphics often lie by omission, leaving out data srt{icient for 

comparisons. The principle: 

Graphics mrrst not quote data out of context. 

Nearly all the important quesdons are left uranswered by tlis 

display: 

19JJ 

codecticut Traffc Deths, 

Before Gq::) ud After (1qi6) 

Strictd xnforcement by rhe Polie 

Against C$ Ex@cding Speed linit 

A few more data points :dd immensely to the account: 

125 Connaticut Tnfic Ded4 

1951 1959 

19Jt 1959

Imagine the very dif,erent interpretitions other possible timepaths 

surrounding the r955-r93e change would have: 

/\/\/\/\ 

Comparisons with adjacent states give a still better conrext, revealing 

it was not only Connecticut that enjoyed a decline in trafiic 

fatalities in the year ofthe crackdown on speeding: 

i, 

Tnfiic Deaths per roo,ooo 

Persons in Comectidt, 

Msacluset*, Rho& hland, 

&d New York, 19J1 r9r9 

GNAPHICAL INTEGIIIY 75 

DoMld T. c,nPbeu ud H. L3umc 

Ros, "The Colmcltidt Ceckdom on 

Speding: Time Series D,rr in Q@si- 

E\perineMl Analysis," in ldwad R. 

Tnfte. eC,., me Q"mtxative Analy\it of 

So./l Prorldr (Rsding, Mas., 1970),

76 cRA?IlrCAI- ?RACTTCT 

Conclusion 

Lying graphics cheapen rhe graphical an everywhere. Since the 

lics ofren show up in news reports, millions ofimages are printed. 

When a chart on television 1ies, it lies tens of millions of times 

over; when a N?1, Yor! Trraes chart lics, it lics goo,ooo times over 

to a great nany important and influendal readers. The lies arc told 

about the major issues ofpublic policy-the government budgct, 

medical care, prices, and fuel economy standards, for example. 

The lies are systematic and quite predictable, nearly always exaggerating 

rhe rate of recent change. 

The main defense of the lying graphic is . . . "Well, at least it 

was approximately correct, wc wcrcjust t(ying to show the general 

dire*ion ofchange." But many of the deceltive displays we 

saw in rhis chapter involved fifteenfold lies, too large to be described 

as approxnnately correct. And in several cases the graphics 

were nor even approximately correct by the most lax ofstandards, 

since they falsfied the real news in the data. It is the special character 

ofnumbers that they have a magnitude as well as an order; 

nurnbers measure 4lartlty. Graphics can display the quantitative 

size of changes as well as their direction. The standard ofgcning 

only rhe direction and nor rhe magnitude right is the philosophy 

that infbrms the Pravda School ofordinal craphics. There, every 

charr has a crystal clear direction coupled with fantasy magnitudes. 

Po.r npoAyhrrx nporrsmfuo(r U9l2 i:ll. Pr1!dd, M^t 24, 1982,p,2,

A second defense ofthe lying graphic is thaq although the desigl 

irselflies, the accual numbers are printed on the graphic for 

those picky folks who want to know the correct size ofthe ef'ects 

displayed. It n as ifnot lying in one placejustifred Iifteenfold lies 

elsewhere. Fcw writers would work under such a modest standard 

ofintegrity, and graphic designe$ should nor either. 

craphical integriry is more lilely to result ifthese six principles 

are followed : 

The representation ofnumbers, as physically measured on the 

surface ofthe graphic itself, should be directly proportional to 

thc numerical quantities represented. 

Clear, detailed, and thorough labcling shouid be used to deGat 

graphical distorrion and ambiguity. Write out explanations of 

the data on the gra?hic iaelf. Label irnportant evena in the data. 

Show data variation, nor design variarion. 

In time-scries displays ofmoney, dedated and stendardized unirs 

ofmonetary measuremcnt are neady always better than nominal 

units. 

The number of information-carrying (variable) dimensions 

depicted should not exceed the number ofdimensions in the 

data. 

Graphics must not quote dara out ofconrext. 

GRA?IIICATINT!GRITY 77

Sources of Graphical Integrity and Sophistication 

Why do artists draw graphi-qs drat lie? WWhy do rhe world's ma.jor 

newspapers and magazines publish them?l 

Although bias and stereoryping are the origin ofmore than a 

few graphical distortions, the primary causer ofinept graphical 

work are to be found in the skills, attitudes, and organizational 

structurc prevailing among those who design and edit statistical 

graPnrcs. 

LacL ofQuantitative Skills of Professional Artists 

Lurking belind the inept graphic is a lack ofjudgmenr about quanrirative 

evidence. Nearly all those who produce graphics for mas 

publication are trained exclusively in the ine arts and have had 

little experiencc wirh the analysis of data. Such experience is essential 

for achieving precision and grace in the presence of statistics, 

but everr textbooks ofgraphical design are silent on how to think 

about numbers. Illustrators too often see their work as an exclusively 

artistic e.ntef,prise-the words "creative," "concept," and 

'\ryle" 

combine regularly in all posible pernutations, a Big Think 

jargon for the sma1l task of constructing a time-series a few data 

points long. Those who get ahead are those who beaurify data, 

never mind statistical integrity. 

The Doccine That Statistical Data Are Boring 

Inept graphics also fourish because many graphic arcists believe 

rhat statistics are boring and tedious. It then follows that decorated 

graphics must pep up, animate, and a1l too often exaggerate whar 

evidence there is in rhe data. For example: 

. Tirre's fint full-time chart specialist, an art-sdrool graduate, 

says that in his work, "The challenge is to present statistics aj 

a visual idea rather than a tedious parade ofnumben.", 

. The opening sentence of the chapter on statistical char* in Jan 

\\ite's Graphn Idea Noteloo&: "\Vhy are statistics so boring?" 

Sample illustrations supposedly reveal "Dry starisrics tumed 

r "tu 

n diffi.dt to l owwhyrhBeeme 

etors ar€ bcing repeated. in Playfair's 

original work these kinds of mndes 

were tror hade; mor€ov€r, th€se cdo 

were nor as wides!.ead in the 193os as 

thet are now. Perhaps ihe reson is .n 

inqese in th. pdceived tred for en!!s 

. , , without a coffomitaht increase in 

thibing in !hei. coBruction- Evidence 

gathded by tbe commitre on ehphics 

oa the Amdi.d sr,ri$iol Asociation 

indietes that fornal $aining in gnphic 

lr$dtnion bas had, marked d(line 

ai at1 lev.h ofedlcation over the lasr 

-Mat_ 

few d.-.ad€s." Howard wair€r, 

ing Newspa!€r cnlhs Fit to Pritrt," in 

Paul A- Kol€a, et al., eds. Pida$ihg o/ 

,/i{iile rd,gd!.2 (New Yo.!, r93o), 

2 Tine. Febtury r!, 1980, p- 3-

80 caA?HIcar PRAcrrct 

into symbolic graphici' and "Plain statistics embellishcd or 

humanized wirh pictures."3 

. Alinebook on graphics,Herdeg's Gnphis f Diagrans,isdcscribed 

by its publisher: "An intemational review demonstrating convincingly 

that statistical and diagrammatic graphics do not 

necesarily have to be dull."a 

The doctrine ofboring dara serves political ends, helping to 

advance ccrtain interests over odrers in bureaucmtic struggles for 

control ofa publication's resources. For ifthe numbers are dull 

dull dull, then an artist, indced many artists, indeed an Art Department 

and an Art Director are required to animate the data, 

lest the eyes ofthe audience glaze over. Thus the doctrine encourages 

placing data graphics under control of artists rather than in 

the hands of those who write the words and Lnow the substancc. 

As the art bureaucracy grows, style replaces content. And the word 

people, having lost space in the publication to data decorators, 

console thcmselves wirh thoughts thar statistics are really rather 

lf the statistics are boring, then you'vc got the wrong numbers. 

Finding the right numbers requircs as much specialized skillstatistical 

skill-and hard work as creating a beautiful design or 

covering a comPlex news slory. 

The Doctrine That Graphics Are only for the 

Unsophisticated Reader 

Many believe that graphical displays should divet and entertain 

rhose in the audience who 6nd the words in the text too dificult. 

For example: 

. Consuner Reports describes the design of their new consumer 

magazine for children: "For the 6rst test issue, CUi professional 

stafproduced an article about sugar that was longer on 

graphics dran on information. We had Gared children might be 

overwhelmed by too many facrs."s 

. An art ditector with overall responsibility for the design of 

some 3,ooo data graphics each year (yieiding 2.5 billion printed 

inages) said that graphics ale irtended more to lure the reader's 

attention away from the advertising than to explain the news 

in any detail. "Unlile the advertisements," he said, "at least 

we dont put naked women in our graphics."6 

tl^n \ . \tt\ne, Ctuphn Il1ed Notebaok 

(Ncw York, 1980), P!. 1a3, 16i- 

awaher Hedea, ed., qathitlDiaCtud 

@tticlr. re76). 

sco,emd R.?r,J, 4j 0u1y t98o), 4o8. 

6louis Silvdstein, Gnphics ar rhe Nd, 

Yorl ?ider," lresenbiion at the Fnst 

ccneral coddence on Social Ghphi6, 

Lecsbug, Vnginia, Odober 21, 1978.

. A news director at a national television nerwork said that 

graphics must be instandy u:rdentandable: "Ifyou have to 

explain it, don't us€ it."7 

This kind ofgraphical thinking lead-r to 

Th€ Compsy Car€t6rt. oae uaerl by I Our d tO 

The Consequences 

Emdoeees dutnE the Fisc.l Y.d tg49 

What E. B. White said ofwriting is also true ofstatistical graphics: 

''No 

one c:n wrire decendy who is distrr.rstful of the reader's 

intelligence, or whose attitude is parronizing."3 Contempt fot 

graphics and their audience, along with the lack of quantitative 

skills among illustrators, has deadiy consequenca for graphical 

work: over-decorated and simplistic designs, tiny data ses, and 

big lies. 

Like censo*hip, tlese constraints on graphical design lead to 

elliptical aad eccentric commuaication. In seeking to avoid 

the subdeties ofthe scatterplot, arrists drew up these convoluced 

specimens, forcing bivariate data into a univariate design: 

souRcts oI INT!GRITY AND SOPIIISTICATION 81 

Tlnr€rview vith aurhor, Jlny 1980. 

Mery Eldor Slqr CratirS Srdrltr6 

(New Yort, 19J2), p. j, eho alpre 

lriateiy dGdibcs rhb as u "lmc@$ary 

3ID W ia Stru.I,J., ed !. B. whjre, 

Tr. Elmar d/Stlc (New Yo!k, r9j9), 

Np Yorl Ttr6, Jue 16. r98o, p. ar 3. 

AIm D. Manvel, "Takrion dd Ece 

noDic Gowt\," Tedtion with R.?r.nution 

Ntu]et|d,9 Awe 1990), p. 3.

82 cRAlIIrcAL ?R,{crrc! 

But beyond reviewing a few examples, let us look more systematically 

at the level of graphical sophistication prevailing at 

differenr pub|c,rons. ln order to makc comparisont ,mong a 

variety ofnewspapers, rnagazines, scientfic joumals, and books, 

I have compiled a rough measure of graphical sophistication-the 

share ofa publication's graphics tlnt are /eldrirtlal. Such a design 

Iinl.s rwo or more v:riables but is rot a time-serie! or: map. 

Relational graptrics are.essential to competcnt statistical analysis 

since they confront statements about cauie and effect with evidence, 

showing how one variable afecs another. The design idea 

is a simlle one, although not quite as simple as the bar chart, timeseries 

plot, or data map. Relational graphics have been used since 

1765 rnd are printed billions of times and ways every year; and 

there is evidence that twelve-year-old children understand the 

design.' 

All these graphics count as sophisticated by our hardly 

demanding measureJo 

tl 

lfi 

1E0 

160 

110 

'\20 

100 

'rt - ri 7* 

, eil* 

It* 

1 **Agt )t€l{48tr0e6t/rr5t 

7.9, 

12 

g: I t!t-)t z 

-16 18 20 32 X4% 

22 21 26 2A 30 

#1f,tPfn+ 

The frequency ofuse ofrelational designs was count€d for 

randomly selected issues from 1974 to 1980 of each of r5 news 

publications. A toul of about 4,ooo graphics were examined itr 

sampled issues. Scaling up the observed data by the frequenry and 

circulation ofthe publication hdicates that th€ sampl€ repreffnts a 

population ofz5o to 3oo billion printed graphical images. 

"j 

eclara lrancis BamberCd, hterlrF 

tation ol Gr.phs at the lleh@bry 

Sdtrool Le\e1," Cdthaln Uiiu*nq af 

An e ica E dt car ian dl Re i. dnh M onagtup h s, 

1l (May 19a2). Ad&rioul daa from 

tqtbooks ed smdudted tsts de 

1oA vdiety ofhesues ofcnphial 

inteUig@.e md complexity are lossible 

dd dother, data dmsity, is ditused in 

p"". ot citv r.itrffiJ 

2.8 Feet per Second Faster 

Neu York Tines, Fehtvry 29,1926, 

P 46. 

Isao Sato sd Miyohei Shidolara, N./ 

Politkt tud E@lania (rokro, tn4), p. 

113; rJlpdese ligh school textbook.

SOURCES OF INTEGRITY AND 

Table r shows the raults, ranling the rJ news publications by 

graphical sophistication. Seven of the papcrs, fr oi\ pneda to tl.re 

WaIl Steet Jownal, prodt:ced no relational graphics among rLose 

sampled and usually limited themrelves to time-series. Other papers 

published more advanced graphics: theJapanese ,4:ahi (a masi circulanon 

daily), 4bahata t Red Fhg. a Communist prrty paper 

that appears, from the data, to have employed a sophisticared and 

talerrted graphical designer in,r97e), and Atton Keizai (a, financial 

daily), as well as Da Spiegel and, The hononisr. Although none 

reached the level ofsophistication found in displays ofscientfic 

daa (a random sample ofzzo graphics ftom Saerce t978-t9lohad, 

42 p€rcent ofrelational design), it is clear that some graphical 

intelligence is possible in news work, 3t leasr in Japan and at a few 

European weeklies. 

Table I 

ciaphi@l soplisti@tion, Wdld ptes, 19?4 19so 

,4!a,'dtd ('Red flag ) (Ja!d, d'ily, 

,ardl,t Sli,6!r (Japd, d.ity, 8,0Cr0.000) 

D4 S?i.gel (cmmy, weclly, 1,000,000) 

fie EadmiJr (Briain, weehy, 170,000) 

Nihoi Keizai shibb't (I^p^, dail.l 

findcial paper, 1,700,000) 

ae Mo,/e (lrdch, daly, a.40,000) 

-B,ri,e$ itl.st (U.S., welly, 800,Ooo) 

Ne/ Y"r* "irs (U-s., daiiy, 900,000; 

Suday, 1,500,000) 

Pd,ld (rJssR, .laily, 10,s00.000) 

Fto*Iwtet Allseneke (Gd|lhy, datr, 

300,000) 

fie Tt,.J (Brirain, daily, 400,0qt) 

irl6nt,Jrof PoJl (U.S., dailr, 600,000; 

Suday, 800,000) 

Itma (U.S., wcctly, 4,300,000) 

Die 2.'/ (c€rnmy, wekly, 300,000) 

Wdll siee t Jarnal lU.S., dîly, 2,U]/J,O|NJ) 

Pe.corag€ of sratisri@l 

ghphics based on more 

rhD one qnabb, bur not 

93% 

7.6% 

s.7% 

2.4% 

1.7% 

o.5% 

0.U/o 

0.4% 

0.U/o 

0.0% 

0.0% 

o.o% 

0.s% 

202 

1t9 

454 

342 

297 

L14 

422 

54 

93 

107 

727 

213 

449 

soPHrsTrcATrON 

83

84 GRAPHIcAL PRAClIC! 

Japanese graphical distinction is consistent with that country's 

heavy use ofstatistical teduriques in the workplace and extensive 

quantitanve training, cven in the early years ofschool: 

. . . no nation rank' higher in irs collecrive pa'sion for rtaristics 

hJapan. srarisrics are rhe subjrcr ofr,holiday, local and uarional 

conventions, award' ceremonies and narionwide statistical 

collection and graph-drawing contests. 

"This year," said 

"we 

Yoshiharu Takaha:hi, a Govemment statistician, had 

almost 3o,ooo entiies. Actually, we had 29,836." 

Entries In ihe lchildren sl sunsrical graph contesr were 

'creened rhree uie. b1 judges. who givifirst prize this year 

to the work oflive 7-year-olds. Their graph creation, titled 

"Mom, 

play with us more often," was the result ofa survey 

of3z claismates on the frequenry that mothers play with their 

off.pring and Lhe reasons giren for not dorng so. . . . Orhcr 

childrerl' *orL examined Lhe f,equency of{,mily phone urage 

and correlated the day's temperature with cicada singing.l' 

Note the reletional design of the last cJrildren's graphic mentioned. 

The five U.S. publications examined rank toward the bottom 

of the world lnt, along wlrh Puoda and t tew European papers. 

Note, in Table r, the complete dominance ofnon-relational 

designs ac thc lower-ranled newspapen and magazines. This is 

unfortunate because the relational graphic, unlike the simpler 

designs, is an explanatary graphic- surely a :ratural for news reporting 

and anal)sis. 

The statistical graphics found in collcge and even high sclool 

textbooks are more sophisticared than those in news publications. 

Indeed, grade school children may expcrience a greater density of 

relational graphics than someone who reads only B siless W?ek, 

the New Ya& Tines, Tine, the Wall Street Jounal, and the Washington 

Posr. Tables z and 3 record the graphical sophistication of 

textbooks and ofa varieq' ofstandardized educational tests. 

A comparison bctwccn these data and Table r suggests rhat most 

news publications ou*ide ofJapan operate at a pre-adLrlt levcl of 

intelligence in graphical design.t' 

llAndrew H. Malcoln\ "Data-Loling 

Japme€ Rejoice on Statistics Day, Ne/ 

Iorl TiueJ, October 28, 1977, p, ^-1, 

1,R€ader of nNs publications, ?anidlult 

the elite !re$, have considehble 

educatiomland profssional att.inments, 

wiih tbe rc$hing sblhi.al skills- About 

30 percent of the 1.j million rcades of 

the suday Ne Yd! Tira attended 

coll€ge, according to a r98o liueJ market 

surev. The audidce for st.tisncal 

gaphics is shan6 tn,n mmy insfrtos

Table z 

caphi€l Sophntietion, Coleg€ and HigL schoot Textbook, 

M€dicine ed lutlic h€alrh: 11 eniclg in 

Judith Teu, er a1., Srdrtriai .4 Gridc ro 

th? Uik"ous 

Int/odudia9 to Plr'hot1gy, by Eta6t 

H ged, cr al. 

Gaetul Ch.nbtrt, by LttN. Pulng 

ry'. o/ Edl,, by Edmrd wit on, er,l. 

Weath6, astronoey, engiD€Eing: 7 

dtides iD Tdu, Statlrid: ,1 c,id. 

ComnuiorioD, vork, edu€rio4 

eoDonigr 20 anicl4 in Tmu, 

Stdtisti6: A C\ide to the Ufksogi 

Polidcal Behauiu a.f the Anentun 

Elnlorak, !y William H. Fluigd 

ed Nmcy H, Zinede 

Eedri.J. by Pail Samuebon 

Dehodaq in An.tita, by R&eft A. Datl 

Anencar Coumhdt, by J^rcs Q. WiJsan 

cr.zirr'l .ryi".?Li, by william Masrenon 

Th. Plojert Phrlns Coure, bt rtstd 

Nu Politis @d &ononks,hy Isrc S.to 

ed Miyohei shinolda 0.pe6e) 

Bia@ical Scien.e: An E@lagi@l Aw.oach, 

Biologicd Scio@s Clriculh Study 

TlE A36itatr E@@6y,by P.oy I. 

!$.iotou: The sttldf of Itunn 

R?lddoar,irr, by Ilvetu Tho6 

Neu Ehks dd social S.i6@, by 

Yokichi Yajim, et al. 0apoe) 

Rie o.f &e Andkan Na!ia., by 

I€wiJ Paul Todd ed Male Cuti 

Masqdd's Ak.n.aa Cortuent, rdised. 

ty Wiliu Mccitughu 

Pccentage of st.tistical 

gBlhics basd on more 

ths one mriatle, lur Dor 

82% 

68% 

66% 

4-,t% 

44% 

43% 

42% 

16% 

8% 

o% 

n% 

4a% 

2Vk 

|a% 

591) 

glo 

uh 

V. 

vh 

souRcls oF INTBCRITY AND SO?FTSTICATION 8J 

l7 

g2 

53 

59 

9 

43 

25 

39 

u 

33 

22 

2a 

132 

3 

5 

39 

70

86 cRA?FrcAr ?RAcrrcE 

Table 3 

Ghphi@l Sophiltication, Eduational Tesis 

N,tioMl uiversity dtuce *min,!ions. 

Review materials, Law School AdmissioD 

Test, UDiFd States 1975 

Standaidized tests for sbde shool, hish 

schoot, and colesei United States, 1970s:* 

Arithmctic, mrthm,tics.,lgehra, od 

analytic gmmedyt 21 ie*s 

Social $udics, hi$ory, dd 

govehrnenr; r+ tefs 

cmml ,bllity, 5 tests 

snlhics bAed on nore 

than one mnabb, bur not Nuhb€r of 

a timseiies or a na! smlhics 

I$qo 

67% 

41.% 

2470 

21% 

* Graphics colected ii Janes R. Bdiger, conpiler, Seia&'1 Stantradi2e'| Tat Itetu 

thdt Metrte Abilitt with CftprB (Wxshington, D.C.: Bueau of Soci.l Science 

And so, just as there is a double standard ofinregrity at a good 

many news publications-one {br words, another for graphics 

so there is a double standard ofsophistication. The statistical 

graphics are stupid; $e prose is often serious and sometimes even 

demanding ofexpertise, as can be seen in these sentences from a 

single issue of the Ner Yo&'fines. 

Rc.y.ling pero.lollrrs Dray po$lone rhe day ofreckoning. bur irs e11'e.rs would 

$on bccone intoldable without a steady de!recirtiotr in tleir puchsing 

power. rioxting stes ofexchrnge .mot restore ev€n r sembldce of 

Numerous facets of the performmcc secn decidedly uDfashiomble ifnot 

doMdght eccenFic: the squu+toed iNrunfr$l phasing lnd rhe fteqnot 

plodding tenpos in the arias, the Ronunricized tratm@t ofthe choFles, n\e 

sendous retards at every cadence, the often intrusiydy drboFre c@muo 

improvir.tions dd an inconsistent attitude tovard exprc$ion whnh rdg* 

fron leaving Malterid emthses to mechdical lircnlisn. 

The Cou.t shows tro dgn ofrelr€ting from irs view thar a state sovemmmt 

is protecied by sover€isn immuity ,g,inst .otrr ord{s to lay ietroactive 

dahages for past violations. 

And Dr. Gdi! crahrm, a nedical dnedor wirh Smirhkline Corp., nakes of 

Thohzine, noied thar neuroleptiG produ@ no cuploria, dd are rberefore 

mrikely to be ab$d by patieft wnh a hiliory of drug or alcohol dep@d@ce. 

"They are, ifanything, dyslhorosenic," Dr. Gmnam eid. 

49 

47

Conclusion 

The conditions under which many data graphics are producedthe 

lack ofsubstantive and quantitative skills ofthe illusrraron, 

dislike ofquantitative evidence, and contempt for the intelligence 

of the audience-guarantee graphic mediocrity. These condirions 

engender graphics drat (r) lie; (z) employ only the simplest designs, 

often unstandardized tine-series based on a small handful of 

dara points; and (3) miss the real news actually in the data. 

It wastes the tremendous conrmrmicativc power ofgraphics to 

use them merely to decorate a few numbers. Moreover, much of 

the world these days is observed and assessed qurtrtiratively-and 

well-designed graphics are far more effective than worib in 

showing such observations. 

How can graphic mediocrity be remedied? 

Surely there is something to be said for rejecting once:nd for 

all the doctrincs that dara graphics are for the unintelligent and 

thar.ror.rics are bonng. These docnines blame the vrcoms lrhe 

audience and rhe data) rarher than the perperrators. 

Graphical competence demands three quite diferent skilis: the 

substantive, statistical, and a*istic. Yet now most graphical work, 

pafticularly at news publications, is under the diredion ofbut a 

single expertisc-the artistic. Allowing artisr-illustrators to control 

the design and content of statisticri graphics is almost like allowing 

typogmphen to control the content, style, and editing ofprose. 

Substantive and quantitative expcrrise must also parricipate in the 

design of data graphics, at leasr if staiistical integrity and graphical 

sophistication are to be achieved. 

SOURClS O! INT!GRITY AND soPIIrsTrcaTIoN 87

PART II 

Theory of Data Graphics

Evuyone spoke of an inJotnatiofl oeEioad, but uhdt theft uas in Jai 

was a on-iflfotuatiotl otte oail, 

Richa.d saui wur@, whdt-If, Co{A-Be (Philalelptt , re76)

4 Data-Ink and Craphical Redesign 

Data gtaphics should drew rhe vieweri attention to the sense and 

substance ofthe dao, not to something else. The data graphical 

form should present the quanticative contents. Occasionally artful_ 

ness of design males a graphic worthy of the Museum of Modem 

A.t, but essenrielly staristical graphics are instruments to help 

peopie reason about quanritative information. 

Playfairt very first charc devoted too much oftheir ink to 

graphical appararus, wirh elaboraregrid lines anddetailed labels. This 

tme-series, engraved in Augusr 1785, is Gom the early paga of 

The Connetcial and Political Atlas: 

CzA.Er {Lu?onrs a2?z 

Ext"ozzs gf INGLAND at ,t rd/ran aZ,Mon?E LaLaru 

,fran d! y.a/ r'/7o at /782 4/ p, "/g/,.ti 

?u.Bddzn tr;u 6 Arila/ ;z.tr l;n/! tlr 4ib-nad tr.:u ,.b EmtnED Tzo.rJ,1*.D ronrr! 

zrttbtd * *t;d aa,4a! ryr.

92 TIIEORY OF DATA CRA?IJICS 

Within a year Playfair had eliminated much of the noniata 

detaii in favor ofcleaner design that focused attention on dre 

time-series itself He then began worLing wirh a new engraver 

and war soon producing clear and elegant displays: 

no.po.r6 sd ImFons ro std Aon DEI{MARI( & NOICIVAY fi.@ v@ 

This improvement ir gnphical design illustrates the fimdemental 

principle of good statistical graphics: 

Above all else show the data. 

The principle is the baris for a theory ofdau graplucs. 

th. nAb h@td L4. ittu ZlSooo a.]z 

1o rzaq

Data-Ink 

A large share ofink on a graphic should present dara-information, 

the ink changing as the daa change. Data-ink is the non-era$ble 

core of a graphic, the non-redundant ink arranged in rerponse 

to variation in the numbe.s represented. Then, 

Data-ink ratio - 

&ta-ink 

total ibl used to print the graphic 

: proponion ofa graphici ink devoted to the 

non-redundant display of data-information 

: r.o - proportion of a graphic that can be erased 

without loss of data-tuformarion. 

A Gw graphics use every drop oftheir ink to convey measured 

quantities. Nothing can be erased without losing informarior in 

there continuous eight tracks ofan electroencephalogram. The data 

change from background activiry to a series ofpolyspiLe busts. 

Note the scale in the bottom block, lower right: 

DATA-INT 93 

I(etueiJr A. Kcoi, FMlanentals oJ EleF 

ttuc$e alogturhy g.tev Y6k, t9j1),

94 TIITORY OT D,{T,{ GNA}IIICS 

Most of the ink in this graphic is data-ink (rhe dots and labels 

on the diagonal), with perhaps ro-zo percenr non-data-ink 

(the grid ticl6 and the frame): 

tu-;. 

E *q-i' 

'E** 

.,' Y . 

*1'.f":i: 

*- 

.:";.;"i*r'"" n"** -' 

*uT'.'.i?,F..." 

[e169dor3'ol4im4$.lidcorFFduc 

Pbldo'a1o@j6.nc 

In this display with nearly all its inl devoted to mattcrs other 

than data, the grid sea overwhelms the numbers (the faint points 

scattered about the diagonal): 

Jotn \ylet Bffie\ size abd crle: An 

Esdf ofl th. Sttectur .f Bialosr (PtiIce

Anorher p&lished vetion of the same data drcve the shate of 

daa-ink up to about o.7 an improvement: 

R.r{ioaship ot Acrut Racs of Rccinnrion io predictcd Ra!6 oo4 cities t960). 

But a third reprint publication ofthe same figure forgot to plot 

the points and simply retraced the grid lines Gom the original, 

including the exces strip ofgrid along the top aad right margins. 

The resulting frgure achievc a graphical absolute zero, a null dataink 

ratio: 

DATA-INK 9J 

The thre smphics were lullisbed iD, 

resFctively, Stanley Kelley, J.., Richud 

l A)rs, ed willi.n c. Bowen, 'RegisthrioD 

md voring: Putting Firsr 

mings list," An{nM Politl.al Sderc 

Re,i@, 61 G96?),371; th@ ieprint€d 

in Edw,rd R. Tufie, ed-, fte Qdrirr'tive 

halysis of S'ridl I'toblem! Frudils, 

M.$. 1e7o), p. 262; md repiinred 

€ain in Wiliam J. Crofty, ed., P,,rlir 

opiniar ed Politi6: A Redlt F.leq

96 TgroRY OF DATA CRA?HrCS 

Maxibiziag the Share ofData-IDk 

The larger the share ofa graphic! ink devoted to data, the better 

(other relevant manen being equal): 

Maximize the data-ink ratio, wirhin reason. 

Every bit ofink on a graphic require.s a reason. And nearly always 

that reason should be that the ink presents new information. 

The principle has a great inany consequences for graphical editing 

and design. The principle makes good sense and generates reason' 

able graphical advice-for perhaps two-thitds ofall statistical 

graphics. For the others, the ratio is ill-defined or isjurt not appropriate. 

Most important, howcver, is that other principles bearing on 

graphical design follow from rhe idea of maximizing the shrre of 

data-ink. 

Two Erasing Pdnciples 

The othcr side ofincreasing the proportion of data-ink is an 

erasing principle: 

Erase non-dara-ink, within rceson. 

Ink that fails to depict statistic.l information does not have much 

interest to the viewer ofa graphic; in fact, sometimes such nondata-ink 

clutters up the data, es in the case ofa a\i& mesh ofgrid 

lines. While it is true that this boring ink sometimes helps set the 

stage for the dara action, it is surprising, as we shall see in Chapter 

7, how often the datr themselves can serve ar their own stage. 

Redundant data-irk depicts the same number over alrd over. The 

labeled, shaded bar of dre bar chart, for example, 

unambiguoudy locates the altitude in six separate ways (any five 

of dre six can be erased and the sixth will still indicate the height): 

as the (r) height of the left line, (z) height ofshading, (r) height 

ofright line, (+) position oftop ho;zontal line, (5) position (not 

conrent) ofnumber at bart top, a]ld (6) th€ number itself. That is

more ways than are nccded. Gratuitous dccoration and reinlbrcernent 

ofthe data neasurcs qenerare much redundanr data-ink: 

Bilatcral symmetry ofdata measures also creatcs redundancy, 

as in the box ploq thc open bar, :nd Chernofffaces: 

@ n @ 

t\ 

Half-faces carry the same information as full faces. Halves may 

be easict to sort (by matching the right half ofan unsorted facc 

to thc lcft halfofa sorred face) than full f:rces. Or clse an 

asymmetrical full face can be uscd to repoft addirional variables.l 

Bilateral symmetry doublcs the space consumed by rhe design 

in a graphic, without adding new in(brmation. The fi:w studies 

done on the perception of symmetrical dcsigns indicatc that "whcn 

looking at a vase, for instance, a su\ect would examine one ofits 

symmctric halves, glance at the other halfand, sccing drat it was 

identical, cease his explorarions. . . . The enjoymenr of symmetry 

. . . lies not with rhe physical properties ofthc {igure. At lcast eye 

movements suggest anything bur symmdry, balance, or rest,"2 

DATA-IN]l 97 

1Bemh.rd nu.r a Hos Riedvyl, 

''Cnphic.l 

Ret'.esenration of Multi 

vrrirtc Data bt Meals ofAsyDnl-trical 

F{es," lautnat oI the Antntu sl4njtndt 

,lso.iar'of, 76 (DeceDrber 1s81), 7J7 

,Leonrrd Zrsne, yisual P{ettidk o.f 

Fd (l.{ew Yo.k, 1970), 

!!. 2j6 2j7.

g8 TIIEOAY Or DArA GRA?IIrCS 

Redundancy, upon occasion, has its uses: giving a context and 

order to complexity, facilihting comperisons over various par* of 

the data, perhaps creating an acsth€tic balance. In cyclical timeseries, 

for example, parts of the cycle should be repeated so that 

dre eye can track any pan ofthe cycle without having to jump 

bacL to the beginning. Such rcdundancy possibly inproves Marey's 

r88o train schedule. Those pcopJe leaving Pads or Lyon in the 

cvening Iind that the! trains run of,'the right-hand edge ofthe 

chart, to be picked up on the left agin: 

Attaching an extra half rycle makes every train in the first 24 

hours of the schedule a continuous line (as would mourting the 

original on a cylinder):

And, similarly, instead of once atound the world in rhis display 

ofsurface ocean current", one rnd rwo-thirds rimes around i. bimer: 

DATA-INK 99 

Kirk Brym ud Michael D. Cof,, Tle 

Cncubtion ofthc Wo.ld oc€dr: A 

Nundicrl Study. Prt 1, A Homogenous 

Modet," Jaundt aJ P\skal oeeaE

1OO THEORY OI DATA CRA?IIICS 

Most data representetloff, however, are ofa single, uncomplicated 

number, and little graphical repetition is needed. unles redundanry 

has a distincdy worthy purpose, the second ensing prhciple applies: 

Fra.e redundanL dau-ink, within rea.on. 

Application of the Principles in Editing and Redesign 

Ju:t as a good editor ofprose rudJexly prunes out unncccssary words, 

so a designer ofstatistical graphics should prune out ink rhat fails 

to present fresh drta-information. Although nothing can replace 

a good graphical idca applied to an interesting set ofnumbers, 

ediring arrd revision are as essential to sound graphical design work 

as they are to writing. T. S. Elior emphasized thc "capital importance 

ofcriticisrn in the work ofcreation irself. Probably, indccd, 

the larger part ofthe labour ofan author in composing his work 

is critical labour; the labour ofsifting, combining, constructing, 

expunging, corrcciog, testing: this frightful toil is as much criticai 

Consider this display, which comparcs cach long bar with the 

adjacent short bar to show the viewer that, urder the vario 

experimental conditions, the long bar is long€r: 

rT. S. Eliot, The F@dim ofcriticlsm, 

in &1e.te'1 E$ayt ry+1q2 

Olew York, 1932), !.18. 

Jan6 T. KuaicH dd N. Bnce Mccurchcon, 

'croseEnltmccndt 

of the 

Sour T,se oD Sinele HutrD Taste 

Papil\ê." Jorndl of Eapdineatal Psr 

thology: Ceaddl, 1os (1979), 76.

Vigorous pruning improves rhe graphic immensely, while still 

retaining all the data of the original. It is remarkable that erasinq 

alone can work such a ransformation: 

"ll,'r 

;L ! L] L] 

''lrl 

LlLtjb 

-lh 

:' 

ILr,,n- 

The horizontals indicate the paired compadsons and would change 

ifthc experimental design changed-so they counr ar informationcarrying. 

AII the asterisks are out since every paired comparison 

was statistically signifrcant, a point that the caption can note. Here 

is the mix ofnon-data-inL and redundant data-ink thatwas erased. 

about 65 percent ofthe original: 

-lr 

.I 

'll r 

lli' 

li 

tl 

f:Ir 

LLL 

LL, 

DATA-INK 1O1

]02 TII!ORY OI DATA GRA?IiICS 

The drta graphical arithmetic looks like this the origiral design 

equals the erased part plus the good part: 

,Ll I'ljrlrl 

'll'[ I']t'l 

'"' 

I r 

LLLI 

" l l ' l 

:llrllll 

ltuLil 

' l 1 r 

LJ L Ll lr 

,:11 

'". 

1,bLt- 

The nexr graphic, drawn by the distinguished science illustrator 

Rogcr Hayward, shows the periodicity of properties of chemical 

elements, exemplified by atomic volume as a function of atomic 

number. The data-ink ratio is les than o.6, lowered because the 

76 data points and the reGrence curves are obscured by rhe 63 darL 

grid marks arrayed over rhe data plane like a precision marching 

band of 63 mosquitoes: 

E ' 

_l 

E 

50 70 

Atomic Number 

i 

i 

+ 

Linu Pauh€, G?ntrdl Clenntry (Sm 

FrDcisco,194?), p- 64.

,ilY.1-:.5::-"'" :rth rhe csenrial informarion of the gr:phic. 

rhe penods and the cmpirical obr.*"rlonr. 

":Tî.,11.."'. rnarks 2nd p111 qfrh. fiamc can be safeiy erased, 

removed trom cbe denomiraror of Lhe d,ata_rn_k rado: 

Tlc-unclurtered displ:y bring. our:norher aspecr or rhe data: 

lii iiii "'il*.-:". do no, fir rhe smoodi ,r,_*".,, .*"., 

llrthei w.ell. The dJrJ_jnl rario has:ncreased 

only 

,. "U.u,.q, 

lhe frame 

;;ri 

ljnes rernarjung a! pur€ non_rDtormrlion: 

I 

DATA-INI{ 1OJ

1O4 TEEORY Ol DATA GRA?HrCS 

The reGrence curves prove essential for organizing the date to 

show the periodicity. The curves create a srmcturc, giving an 

ordering, a hierarchy, to the flow of infomation from the Page: 

d 9 

F- 

I 

Jg 

E 

60 70 

Atomic Numb€r 

Restoring the grid fails co orgadze the data. The ticks are too 

powerful, and drey also add a disconcerting visual vibration to the 

graphic. With the ticks, the reGrence curves become all the more 

necessary, since the eye needs some guidance through the rneze of 

dots and crosses: 

E ' 

f 

SB 

E

The space opened up by erasing can be effectively used. Labels for 

the initial elemens ofeach period, an alkali, show the beginning 

ofeach cycle in the periodic table of elements-and in the graphic. 

The unusual rare-earths are indicated. In addition, the label and 

numbers on the veftical axis are mmed to read Gom left to rieht 

raLher *un boLrom to rop. maLirg the graphic slightly more 

accessible, a linle more friendly: 

Conclusion 

I 

I 

Na ll ! 

"i ii 

' l r I 

, r 

,, | 

.,J 

.i lY 1 ; 

I 

I 

Five principles in the theory ofdata graphics produce substantial 

changes in graphical design. The principles apply to many graphics 

and yield a series ofdesign options through cycles ofgraphical 

revision and editing. 

Above all else show the data. 

Maximize the data-ink ratio. 

Erase non-data-ink. 

Erase redundant data-inL. 

Revise and edit. 

I 

DATA-rNK 1O5

With sauge piawes jll then gaps 

And o'er u abitable dowtrs 

Plate ebphants for want of touns. 

Jonatlo SwifCs indictnent of rTth

5 Chartjwnk: Vibrations, Crids, and Ducks 

The interior decoration ofgraphics generates a lot ofinl that does 

not tell the viewet anyrhing new. The purpose ofdecoration varies 

to make the graphic appear more scienti{ic and precise, to enliven 

the display, to give the designer an opportunity to exercise artistic 

slills. Regardles ofis cause, it is ail non-&ta-ink or redundant 

data-ink, and it is often charlunk. craphical decoration, which 

prospers in technical publications ar well as in commercial and 

media graphics, comes cheaper than the hard work required to 

produce intriguing mrmbers and secure evidence. 

Sometimes rhe decoration is thought to refect rhe artist's fundamental 

dcsign contribution, capturing rhe essential spirit ofthe 

data and so on. Thus principles ofartistic integriry and creativiry 

are invoked to defend-even to advance-the cause ofcharqjunk. 

There are bctter ways to portray spirirs and essences than to get 

rhem rl'rangled up wrrh sLari.rical graphics. 

Fortunately most char{unl does not involve artistic considerations. 

It is simply conventional graphical paraphemalia rourinely 

added to every display that passes by: over-busy gdd lines and 

excess ticks, reduadant representations ofthe sinplest data, the 

debris ofcomputer plottiag, aad many ofthe devices generaring 

derign variation. 

Like weeds, many varieties ofcharlunl flourish. Here three 

widespread types found in scicntific and technical research 

work are catalogued unintentional optical art, the dreaded grid, 

and thc self-promoting graphical duck. A hundred charlunLy 

examples from commerciai and media graphics have been forgone 

so as to demonstrate the relevance ofthe critique to the professional 

scientific production ofdatr graphics 

Unintentioml Optical Art 

Contemporary optical art relies on moird effects, in which the 

design interacts with the physiological tremor ofthe eye to produce 

the distracting appearance ofvibration and movemmt.

1O8 TITEORY OF DATA GRAPIiICS 

The efect extcnds beyond the ink of the design to the lvhole Page. 

When exploited by the experts, such as Bridget Riley and Victor 

Vasarcly, op art effects are undoubtedly eye-catching. 

But statistical graphics are also oftcn drawn up so as to shimmer. 

This moiri vibration, probably the most common form ofgraphical 

clumer, rs inevitabiy bad art rnd bad data graphics The noise 

clouds the flow ofnrformation as these examples from teclurical and 

scientiic publicarions illustrate : 

[--5 

4u.fu"fii;il 

n::#g:'ss 

TECIDOS DTALGODAO 

( coToNNADES ) 

E6N\\\ 

NNN 

lliluk kffi-ffi 

iir- -- rii 

- 

( COITON IEXTILTs ] 

In$iruto de Expdsso Commcrcirl, 

Btd\il: Ctdphnat E.anoni.avEndtnti1s 

(Rio dcJeeno, 1929), p.15- 

@! 

L ] 

i;ndm trc;* 1 

pR00uciAlrDEricrD05 

ffiffiff 

t( /ip(rRwD"'l 

I TmPoRTAQA! | 

ffitutu-ffih -m 

-

t I 1.5 

TE 

.9 .^ 

I 

0.0 

On chis page, what should have been simple tables are tumed into 

bad graphics pubtished in major scienrifc journals. Above a duck 

moir6 with an unintentional Necker Illusion. as the two back olanes 

optically flip to the front. Some prrrmids concerl others; and one 

va.iable (stacked depth ofthe stupid pyramids) has no labet or scale. 

Below, we learn very litrte about data, but do discov€r that moir6 

vibration may well be at a ma-rimum for equally spaced bars: 

CIIARTJ UNK 

Months after Operation 

Figur€ 2. Senal Echocardiogrsphic ass€ssmenrs ot rhe seve y ot Reguqirarion in he puhonary Aitoglaff in 3t pari€nrs. 

Th6 numedcal gradss werc assigned according ro iho sovsrfty ot rcgurgitarion, as toltovrs: o, nono; O.g, rdv6r: 1.0 !o 1.5, mitd:2_0, 

moderato: and 3.0, ssv€re, 

:l 

s 

! 

: 10 

Nichola T. Koucholtos, er ai., 

''Replaemdr ofrhe Aorti. Roor with 

a Pulmonary Autograft in Chndrm od 

Youg Addts with Aonic-Valye Disqsq ' 

me Ntu EntlardJastual a.f Medkine, 3p 

J.he, T. Knicki dd N. Bre 

Cutcheon, 

Me 

'CroleEnllu@mdt 

of rhc 

Sou T.ste on Single H1lmD Tasb 

Pepinae," Junat 

.tbtogf: Cd-al, 

oJ E lennent.t P\yros 

(le?s). j6- 

Eain M. CoDford md Maii€ n. Huot, 

''Glucose Thsfer from Male to Female 

Scbistoson6," S.idc, 2r3 (Septdbd

110 TIITORY OI DATA GRAP1IICS 

And, finaliy, from the stylc sheet once providedby the Joumal 

of the Anetican Staritical lssoc;ation, a graphic described as 

"an 

example ofa 6gurc prepared in rhe proper form : 

A. Avetage Prcbabilltles ol W lton Nl1.1l 

wnhn=10 

Ar'ERAGE PROSABILITY 

The display required r3r line-strokes and r5 digits to cominunicate 

its simple informrtion. The vibrating lines are poorly drawn, 

unevenly spaced, and misaligned with the vertical axis. 

Vibradng char{unk even frequents the graphics of major 

scientiic j oumals: 

Tle t@ most lrequdtly citcd 

(fooEbted) scidiifrc jomal: rsdom 

enple ofissrs lublishcd 1980 1982 

Jo,ndt 

6.f Blolosicdl Cheflnty 

Jounat aJ the Ahdldr Chehnal sochtr 

Joutuat of chenndt Pltt'i6 

Bio.hini'a et Biophtsi.a Add 

Proreedihgt oJ the Ndtiondl Auden! .J 

Neu Engtdnd Jottual o_f Medlci"e 

2% 

2% 

3% 

a% 

t't% 

12% 

t5% 

21% 

568 

565 

327 

432 

225 

438 

364 

3tt 

338 

"JASA 

sryte sheet," Joutudt o.f ttE Anonan 

Stdtktudl Asociatiai. 7! (Maeb

Moir€ effects have proliferated with computer graphics (in 

programs such as Excel). Such unfortunate panerns were once 

generated by means ofthin plastic tranGr sheers; now the 

computer produces instant chartjurL. Shown here are a few of 

the many vibrating possibilities. Crosrhatching should be replaced 

with tint screens ofshades ofgray. Specfic areas on a graphic 

should be labeled with wot& rather than encoded with hatchine. 

l:31:!:!f:l a a a a o t r'r'o'o'r'o'o'r'r.r 

a a a a a ................... 

NNNNNNNNW NRN./////////t'/m 

ffiHffi liiiiiffi:iiiiiiiiiii i:i::iii::iiiliii i:iii:iiiii: :::::::::::::: 

))))))))))))})Dl))l))) : 

CIIARTJI'NR 111 

a a a a a o a a 

a a a a a a a ( 

ryfiw lEii*iEiEiE!;: ililililffiililffiililfl il[ilil1t ffi I ffi ililillll 

ililillililtii1iltt 

illtititl 

ffi i11:ii1:i:il1:ii illlllll:l-E ffiffi ilffi 

fil.ill]Tff[ffn i]fidrffiiifitdfiiliitfilfifit Ffiililil i ffi i : : i i i i i i : i i i : i i 

| | I il lllllllll 1ililililrililr ililililililil|ilililt ilililil1ililiii iiiiiiiliiiilit ilititi ililtl 

,m ilililIililIilililtililililililtilil | | | | | I | | | 

ffi ffix* ilmffi ,ttttllltl|ltl1tl||1ttlll >)>>>)))>" 

1+++H# N\N\N &ttt9tt mm$ NSTNSSS \\\'* 

This form of clurlunl is a twentieth-century innovation, and 

computer graphics are multiplying it more than ever. The handbooks 

and textbooks ofstatistical graphics, along with user's manuals 

for computer graphics programs, are 6lled up with vibrating 

graphics, presented as exemplars of design. Note the high

112 TAEONY OF DATA GRAPIIICS 

proportior ofchariunky graphics in the morc recent publicarions. 

Computer graphics are particularly acrive: 

Tertbmtr rd hadbooks of sbtistnal gblhnsi Perentage of 

dd mo,l5 for comluter smlhics programs grxlbics rith Total nuhbet 

(ordmd by date ofpublicarion) mond vjbhdon ofsmDhics 

wilttdc. Bti an, Gaphi. ^4ekatuJt 

A@riry Fdra (New Yorl, 1914) 

R. S.tet, t"r G'd?1t4,6 (Pr.is, 1932) 

Hdb€rr Arlin dd PiyDrond R. Colton, c/a?rJi 

H.'o tn Mdke ad ue m (New Yo'k, 1936) 

Mrry Elemor sped, C/rarmg srdr6,i6 

(New York, 1952) 

Atu c. I\.gea, Ctafin chdn' Handbook 

(W.rhington, D.C , 1961) 

F. J. M@khose and H. R. vllkinson, Ma?r 

dd Dtdj',,J (rondon, rhird €dition. 1971) 

c.rvin F. Schnid dd Stanron E. Sclxnid, 

HMdboak a.f Ctofin Ptes atr, (NewYork, 

A.J- M,ccregor, C/dprnJ Sir?/rried 

The usli ndut for a widdy disrlibured 

conrputer smphic !ac*ase: s.1s/cR 4PH Uei' 

clir'. (c!)'' Nordr cfotin . 1980) 

The nmud for a very er@sive conlurer 

grrhis proerrnr: TeI-.4-c td_l Ute" ' Mafl"al 

(Sd Diego, 1!]81) 

'127a 

29% 

17% 

32% 

14% 

220 

14% 

6a% 

53% 

Can optical art effects ever produce a bener graphic? Bertin 

exhons: "h is the designer's dury to make the rrrosc ofthis variation; 

to obtain the resonance [of moiri vibration] without provoking 

an uncomfortable scnsation: to firt with ambiguiq. without 

succumbing to it."1 But can sratistical graphics successfully 

"flin 

with ambiguiry"? k is a clever idea, but no good examples 

are to be found. The key difiiculty remains: moiri vibration 

s an undisdpliaed znbiguiry, with an illusive, eyc-straining quality 

thar contamimtes the entire graphic. lt has no place in data 

gmphical design. 

The Grid 

One of the more sedate graphical elements, the grid should urually 

be muted or completely suppressed so that its presence is only 

impliot-lest it compete with the data. Grids are mosrly for the 

initial plotting of data at home or ofice rarher than lbr putting 

I 

5 

95 

't34 

399 

459 

I 

J^.q\es Beii'n, S.thialos aJQaphi.s: 

Diagrd',r j Net,,/eJ, MoF (Madison, 

wisconsnr, 198j, rnnrlated by wilian J. 

!ers), !. 80 i this book is rhc Euslidl 

tradationof Bcrtin\importdtwo.k, 

sln ialasie Etu?h4"e (Pa^, 1967) -

into print. Dark grid lines are chariunk. They carry no informrtion, 

clutter up the graphic, and generatc graphic activiry unrelared 

to dara information. This grid camoullages the profile ofthe d:ta in 

the age-sex pyramid of the populadon ofFrance in 1967: 

A revision quiets the grid and gives emphasis to the data: 

cHATTJUNK rrl 

BAed on dali in LTdiur Narional de 1, 

Sradtiqu. et des Erudes Economiqus, 

Aintone ndtudEt de la FrMe, ry68 

earis, 1968), p!. 3a3ji redram in 

Henry S. Shryock ddJacob S. Siesd, 

The Methads md Mateidls oJ Denoga?hy 

(wnhington, D.C., 197r, vol. 1, 24a_

E '.r 

tsT 

114 TIITORY OI DATA GRA?HICS 

Thc space occupied by the doubled grid lines consumes 18 percent 

of the area of this otherwise most ingenious design, a "multiwindow 

plot." Optical white dots appear at the intersections of 

the gnd lnes. (The plot shows the following: The large square 

contains X1, X, scatterplo* for the indicated levels ofx, and X,. 

The marginal plots on dre right are conditioned on X, and the 

plots al the top on X,. The upper right comer shows the unconditional 

X,, X, scatter.) Redrawing eliminates the vibration: 

3 t.i I 

trtrHHtr' 

PLOT OF PARTICLE PHYSICS MOMENTL]M DATA 

- 11 

ffi -GrT; 

l,ritl 

r_l 

f.7;:al 

Ell 

r--.-:-l 

E] 

t-tl 

l.+! 

t.r 

|'1': I 

ft 

ti:4ti I 

Paut A. Tlrey and John W. Tlkey, 

"Data-DrivenView 

Seletioni Agglom' 

eration dd Sharp@jng, ' id Vic Bamett, 

eC.., Iflteryrering Mthi"dndft Datd \Chichester, 

Engl,rd, 1981), 231-232. 

ffil 

$i 

:,;l'. 

r;}*

The grid in the classic Marey train schedule is very active: 

Thinning the grid lines helps a little bit: 

CIIARTJUNK 11J

"l!;tr*; 

116 TIIXORY OF DATA CRA?IIICS 

A better treamrent, howerer, is z gray grid: 

When a graphic serves as a look-up able, then a grid rnay help 

in reading alrrd interpolating. But even in this case the gridr shouid 

be muted relative to the dam. A gray grid works well and, with a 

delicate 1ine, may promote more accurate data reconstruchon 

than a dark grid. 

Most ready-made graph paper comes with a darkly printed grid. 

The reverse (unprinted) side should be used, for then the lines 

show through frintly rnd do not clutrer the &ta. Ifthe paper is 

heavily gridded on both sides, throw it oul 

Self-Proneoting Graphics: The Duck 

When a graphic is tahen over by decorative forms or compurer 

debris, when the dara measures and srructures become Design 

Elements, when the overall design purveys Graphical Style rather 

than quantitative infomation, then that graphic may be called a 

lac! in honor ofthe duck-form store, 

"Big Duck." For this building 

the whole structure is itselfdecoration, just as in che duck data 

grzpbrc. In Learning fron Las Vegas, Robert Yenturi, Denise Scoft

Brown, and Steven Izenour writc about the ducks ofmodem 

architecture-and their thoughts are relevant to the design ofdata 

graphics as well: 

When Modem architects righteously abandoned omament on 

buildings, they unconscioudy designed buildings rhat werc 

ornament. In pmmoting Space and Articularion over symbolnm 

and omament, they distorred the wholc building inro 

a duck. They substituted for the innocent and inexpensfue 

practice ofapplied decoration on a conventional shid rhe 

rather cynical and expensive distortion ofprogram and structure 

to promote a duck. . .. k is now rime ro reevaluate the 

oncc-horrifying staremcnr ofJoLn Ruskin that archirecrure is 

the decoration ofconstrucrion, but we should append the 

warning ofPugin: It is:ll righr to decorate consrrucion bur 

never construcr dccoranon-. 

cHARTJUNK 117 

,Robet Verturi, Dcnisc Scoft B.oM, 

.nd SteYer lzenou, I4nirgl'o,, l,r 

/eJdr (Cambridsc, rcvised editi.n, 

1977), p- 16t. The rniti"l sratencnt of 

the duck concept n foud on pp.37 1ol 

a& D!.1, Flandes, New Yorkr phore 

snph by Edward Tuftc, Jury 2o@.

11d THEORY OF DATA GRA?HICS 

The addition ofa fake perspective to the data structure clutters 

many graphics. This variety ofchar{unl, now at high fashion in 

the world of Boutique Data Graphics, abounds in corporate annual 

r€ports, the phony statistical studies presented in advertisements, 

the mas media, and the more muddled sorts ofsocial science 

A series ofweird three-dimensional displays appearing in the 

magezine Anetiean Edurarion in the r97os delighted connoisseurs 

of the graphically preposterous. Here 6ve colon report, almost by 

happenstance, only frve pieces ofdao (since the division within 

each year adds to roo percent). This may well be the worst graphic 

ever to frnd its wav into Drinti

Thcre are rome superbly produccd ducksr 

ffi 

h 

r 

^sEo- 

CHARTJUNK 119 

witlian L. ($.1, et,t., 7t. cduo,,id 

ttdtl trlar (Sa.hnreDto, r973, r979), 

Applied lrigation Water 

19t12 

tttbb 

E^!q 

Eh 

\ 

tE*tu 

Eb 

!b 

tE'4tu 

l@tu 

5.furd 

t+ar"'6aG-

12O TIIEORY OF DATA CiA?IIICS 

Occasionally designers seem to seel credit merely for possessing 

a new tecinology, rather than using it to make beaer designs. 

Computers and their aftliated apparatus can do powedil things 

graphically, in part by turning out the hun&eds ofplots necessary 

for good data analysis. But at least a few computer graphics only 

evoLe the response 

"Isn't it remarkable tJrat the computer can be 

programmed to draw like that?" instead of "My, what interesting 

darz." 

a0 

55 

o5! 

drs 

=r0 

-!5 

3t0 

E2s 

*20 

! rs 

I to 

llrf 5i0il cnliE colf colf 

utEi inqE aovl?or ttrEig 

tlluE Sitis 

@ 

@ 

a 

a 

Z 

N! 

NI 

N 

NI 

lliflAtIOx 

(I.ql5l 

tlxErtlotiE|rl (xrl00l 

3||0irn0Et (L.Ett 

mCt (1.10!l 

CilliE lI-l2ll 

The symptoms of the We-Used-A-Computer-To-Build-A-Du& 

Sy1&ome appear in this display liom a professionaljournal: the 

thin substance; the clotted, crinkly lettering all in upper-case sans 

serif; the poindessly ordered cros-hatching; the labels written 

in computer abbrwiations; the optical vibration-all these the 

by-products ofthe technology ofgraphic fabrication. The overly 

busy vetical scaling shows more percentag€ marLen and labels 

than there are actual &ta points. The observed values ofthe 

percentages should be printed instead. Since the inforrnation conrirts 

ofa few numbers :nd a good m:ny words. it is b€sr ro pass 

up the computerized graphics capability this time aad tell the 

story with a table: 

ColT. ?qIEi (I.lsU 

Colft0EiCE (t.2!ll 

mtEionlE (x-s!?l 

CoI?E!EicE tx-!221 

Anhu I!- Mils, Edic N. Goldenberg, 

ud Lutz E$ring, "Tyleset Politi6: 

Itup.ct of NewspaFa on Plblic Cofrden@," 

Ahqidr Politk|l Stide 

R.!iee, 7 3 (19j9), 67-84.

Contdt dd tone of Iront-page 

articls in 94 U-S. newspapes, 

Odober &d Novernbd, 1974 

Wate.gate: defddnts ed lrosotoB, 5t 

!ord: ladon ofNixon 

Inflarion, ligh cost ofliving 415 

Goverm€nt conpetence: costs, $ality, nn 

elari€s of public chployees 

Co.ndence in govemncnr: powe. of 6 

sp€ial rersrs, t]usr in political 

lqdes, dilhonsry in lolirjd 

GovemeDt !owd: regdation of busine*, 154 

seOecy, control of ciA ed Flr 

Crihe 123 

Sttonages: energy, food 

Conclusion 

103 

10u 

68 

Pdccnt ofarticlc! wirh 

sle.ific pds@ or lolicy 

2a% 

3W 

52% 

4 

30% 

25% 

L3% 

r6% 

Charqjunk does not achieve thc goals ofits propagators. The 

overwhelming fact of data graphics is that they stand or fa1l on 

their content, gracefi:1ly displayed. Graphics do not become 

attractive and intercsting through the addition of omamental 

hatching and false pcrspective to a few bars. Charqjurk can tum 

bores into disasters, but it carl ncver rescue a thin data set. The 

best designs (for examplc, Mimrd on NapolcoD in Russia, Marey's 

graphical train schcdule, the cancer maps, the Ti,res weather hiF 

tory of New York City, the chronicle of the annual adventures 

of the Japancse beetle, the ncw view of rhe gala\ies) arc inbiguing 

and aniosity-proroleing, drawing dre viewer into thc wonder of the 

data, sometimes by narrative power, somctimes by immense detail, 

and sometime-s by elegant prcsentarion of simple but interesting 

data. Iut no infomatiof, no sense of discovcry, no wonder, no 

substance is gencrated by chardunk. 

Forgo ctrarqjunlc, including 

moiri vibration, 

the grid, and the ducL. 

CIIARTJUNK 121

Painting is special, seyrate, a nattel of neditatian and nntenplatioti, 

Jor ne, no physical action ot sorial spot- As nuth onyiausness as 

pssible. Clatity, conpleteness, quintessence, qltiet. Na noke, flo s.hnutz, 

na s(hn ,, o Iauw s&uanefti. PlrIedion, pa*i,rt1pts. cansananc. 

consunt ateness. Na palpitations, no gesticulatiotl, no lrotesqtlerie. 

Sphitraliry, sercnitl, absoluteness, ohefet1ce. No autonatisn, 

fla auideflt, no anxiet/, tro uttharcis, no cha ce. Detachneflt, 

disi te|estedfless, tho ghtfulness, transrcndmce. No hunbuging, 

no bunon-holing- nt e:plo ation. no ni,ing &hgt up. 

Ad Reinlardt, statement 

6r tle caraloguc 

of rhe exhilirioD, "Thc New Decade: 

3J Anericd Painters and Sariptors," Whitfty Mlsfln ofAncrid AJt,

o Data-Ink Maximization and Graphical Design 

So far the principles of maximizing data-ink and erasing have 

helped to generate a series ofchoices in the proces ofgraphical 

revision. This is an imporant result, but can the ideas reach beyond 

the dctails and parriculadtie! ofeditirg? Is it possible to do 

what a theory of graphics is supposed ro do, that is, to derive new 

graphical forms? In this chapter the principles are applied to many 

graphical designs, baric and advanced, including box plots, bar 

charts, histograms, and scatrerplors. New designs result. 

Redesign ofthe Box Plot 

Mary Eleanor Spear's 

"range bar" 

|.-Rong€ 

andJoh Tuleyi "box plot" 

from lo*est to highestomouni---------l 

l* Inierouo.ril€ Ro nEeJ 

Mary llanor S|eu, Cltrttj Stdtntid 

(NewYork,1e52), P.166; and]ohn Ww. 

Tukey, E ?ldfurory Ddia /ralpJ (Reading,

124 THTORY OF DATA CRAIHTCS 

crn be mostly erased withour loss of information: 

The revised desigl, a 4adl,ile p/ot, shows the same live numbers. It 

n casy to draw by lund or computer and, most importandy, can 

replace the conventional scatterplot frame. The straightcdge need 

only be placed on the paper once to draw thc quartile plot, compared 

to six separare placings for the box plor. An altemative is 

but this design will not work cllcctively to ftame a scanerplot. 

Nor does it look very good. 

Pcrlups spccial emph:sis should bc given to the middle half of 

rhe distribution, however, as in thc box plot. This can be done by 

changing line weights 

or. even oc,rer. b1 olt.e*ing r\c niddle \aJf' 

This iarrer design is thc prcfcrtcd form of the quartile plot. It uses 

the ink efi'ectively and looks good. 

tn drese revisions ofthe box plot, the principJe of maximizing 

data-ink has suggested a variety ofdesigns, but the choice of the 

best overall arrangcment naturally also resn on statistical ard 

aesthetic criteria-in other wor&, the procedure is onc ol reaonable 

data-ink mrximizing.

The same logic applies to many similar dcsigns, 

"pa 

such as rhis 

rllcl scle'natic plor. T\cong..,l required 8o seporrrc plac_ 

ing, ofrhe v,aighredge. ;o lrorizonr:1. and 3o rerric-lr: 

F 

ir i. 

ilfl r!r 

|?uIu 

-L- 

i -,- T i 

_i 

' 

l_ 

-I- -f 

An erased version requires only ro verticals to show the same 

in{6lmation: 

I 

The largc reduction in the amouat ofdrawing is relevanc for the 

u.e o[:uch de.ign. rn rnformal. cxntorarory dara analy,is. *here 

lhe re*art-h worker's rime 'fould be de\ oreo lo rrurrers orher 

than drawing lines. 

-! 

DATA-INK MAXITIIIZATION 1.2J

126 TIIIORY O! DATA CRA?HICS 

Redesign of the Bar Chart/Histogarn 

Here is the standard model bar chart, with the design endorsed by 

thc practices and the style sheets of many statistical and screfltfic 

publicarions: 

ks architecture difers little from Playfair s original design: 

&port! hd li+db oI S.oaI/ArlD b ana &no itfleent prs 6. oDe Y.q F(E cld(!e' r.o to chri{hlqr y3a 

rtk ,rg/+ Arv'?,t 2.. ?zn t'hazt/b-,t ?o'./: .a.l z'tu tla.t tua R 27tu d' /lll/'4/bu 1-?'*t 

dk-&ktut*z.r6r8e.d 

/'4^+ 'rz td r-e

The box can be erased: 

And the venical axis, except for th€ ticli: 

Even part of the data measures can be erased, maLing a ulire 

jrid, which shows the coordinate lines morc precisely than ticks 

alone: 

ts% 

1$h 

5% 

DATA-INK MAXll\,llZATION 127

128 TFIORY O! DAIA GRAPIIICS 

The white grid eliminates the tick marks, sircc the numerical labels 

on the vertical are tied direcdy to the white lines: 

t5% 

to% 

5% 

Although the intersection of the dncker bar with the thinner baseline 

creates an attractive visual efect (but also the oPtical illusion 

of gray dots at the intenections), the baseline can be erased since 

rhe bars dc6ne Lhe end-point at the bottom 

I 

ll., 

llllr 

Still, a thin bascline looks good: 

I . 

lr.l 

illlt

Erasing and data-ink maximizing have induced changes in the 

plain old bar drart. The teclniques-no frame, no venical axis, 

no ticks, and thc white grid-apply to other designs: 

a . 

- 

a 

80 

.llj 

10 

Var "b F 1lrdlh N!i.hed Box plot 

I 

I 

T 

!'l I] 

il 

I 

I 

I 

f 

! 

tI 

DATA-INI( MAXIMIZATTON 129 

Rolert Mccirl, Jobn \v. Tder, dd 

Walhe A. Ilren, Vanations ofBox 

Plois,' Anoiuk Sknsticif, 32 (a978),

13O THTORY OF DATA CRA?HICS 

Redesign of the Scatterplot 

Considcr thc standard bivariate scatterplot: 

A useful fact, brought to norice by dre maxnnization and crasing 

principles, is dut the frrme of a graphic can become an effective 

data-communicating element simply by erasing part ofit. The 

frame lines drould extend only to the measured limits ofthe data 

radrer than, as is customary, ro sorne arbitrary point likc thc ncxt 

round nurnber marking ofi'the grid and grid ticks ofthc plot. 

That prrt of the frame exceeding the limits ofthe observed dara 

is nimmed o{I: 

'lhe 

resrlt, a tange-Jtane, explicitly shows the maximum and minimr:m 

ofbothvariables plottcd (a1ong with the range), infbrmation 

available only by extr:polation and visual cstimation in the conventional 

design. The data-inh ratio has incrcased: sorne non-dataink 

has been erased, and dre remainder ofthc frarne, now carrying 

information, has gonc over to the side ofdata-ink.

Nothing but the tails of tha6"-e oeed cl,r.og"' 

D4rA-TNXMAXTMTZ_ATION 131 

Corrientional Scatteplot 

RdgFFlan€ 

A range-Game does nor requLe any vievring or decoding instructions; 

it is not a graphical puzzle and most viewen en easily 

rell whar is going on. Since ir is more tuformarive abour tbe daa 

in a clear and precise nlanner, the range-ftame should rcplace the 

nonJata-bearing fiame in many graphical appticatioos.

132 TIIIORY OI DATA GRA?IIIC5 

A smrll drift in the remaining ink 

a quartile plot: 

turns each range-ftarne 

Irasnrg ald cditing has led ro the drsplay often cxrrx rrmbcrs 

(thc rninnnurn, rnaximum, two quartiles, and thc rncdian for both 

variablcs). TJre design is useful lor anrlytical and cxploratory data 

analysn, as *-eli as for publisired graphcs where sumnary characterizations 

ofthe Darginal distributions have interest. Thc dcsign 

is nearly always bener than the conventionally ftamed scaterplot. 

R.rnge-franes can also present ranges along a singlc dimemion. 

Here the historical high and iow are shown in the vcrticll fleme. 

This is an cxccilcnt practice and should be used widcly in all sorrs 

ofdisphys, both scientiic and unscicntfic:

Finally, the cntire frame can be tumed into data by framing the 

bivariate scattcr widr the marginal distribution ofeach variable. 

The dot-dash-plot restlts.l 

. 

- 

= 

("i,yit 

I tl 

xi 

rr rl I Iii | | | | 

The dot-dash-plot combines the two fundammtai graphical 

designs used in statistical analysis, the marginal frequency distribution 

and tLe bivariate disrribtfion. Do.-dash-plots make routine 

whar good data analysts do already-plotting narginal andjoint 

distributions together. 

An empirical cunulativc distribution ofresiduals on a normal 

gdd shows the ourcl 18 terms plus the 3oth term, with all 60 

points plotred in the marginal distribution: 

DArA-INX MAXTMTZAtTON 133 

I Tne terninology {ollows tradition, for 

ssrerploB w€re oo.c.aled dor diagmDs" 

lor mmple, in R. A. Fishcls 

Statiltitdl Methodt .fot Reyfuh wotkdt 

(ldinbqsh, 1e2J). 

Crthbert DaDiel,,4plrddd's o/sr,t'ti6 

to Indsstlial Exrynnent.tioa lNew \otL, 

r9?6), 

!.1Jj.

134 TIITORY OF DATA GRA?IIICS 

Similarly, this data-rich graphic ofsignals from lulsers shows both 

:: 

lt r.S MHz 

FREOI.]ENCY 

CELL 

, 

Narro$bad sp..tr! ot irdn,idual subPrls$. Xactr poitrr ot rhe irreffiiy 

1"(l) plolted on the risht is the sun ot rhe disiribnrion of itrtensiies acrcFs the 

rcerr€r bud\ridth slioNn jn th. o!nte..,{t tte tor is ptoit€d th. spect.un i\cfused 

or.r tb€ puhe.In thc linit oi mrir rhousxDdr of Jjuls.s this Noutd shof rhe .crcjler 

Tilroih] H. HanIiN rd Bahey J. 

Rickett, "Pdsar SisDl Procesing," in 

B@i Aldei, et al,, edr., M.dodJ i, 

cdhlntdtiahdl Phrsi6, yolune 1a: Rddio 

lr,',,drl G\Iew Yoil,1e7j), p- 1o8.

The fringe ofdashes nr rhe dot-dash-plot can connect a serics of 

bivariate scaters in a /rg?lot (since it raembles a set offringcd 

rugs and covers the statistical ground): 

Reflecting the one-dimcnsional projections ftom each scamei, rhe 

dashcs encouragc rhe eye to notice how each plot filtcn and translates 

the data rhrough thc scatter from one adjacenr plor to the 

ne).t. Sometinet ir is usefui to thinl< ofeach bivariate scaner as the 

imperfect empirical representarion oferl underlying curve that 

transforms one variable into another. In rhe rugplot, the sequence 

ofvariables can wander offas appropriate. The quenritarive hisrory 

of a singlc observation can be traccd through a series ofone- and 

two-cllmenslonal contexb_ 

DATA-INK MAXII4IZAT]ON IJJ

13O TIITORY OI DATA GRA?IIICS 

Conclusion 

The lirst parr ofa theory of data graplncs n in place. The idea, as 

describcd in rhc prcvious thrcc chaprcrs, is that most ofa graphic\ 

ink s|ould vary in re"pon.e to d:rta variarion. l lre theory lr.r. 

sometling to say about a great variery of graplncs workadry 

scientifrc charts, rle unique drrwings ofRoger Hayward, the 

exemplars of graphical handbooks, newspaper displays, computer 

graphics, standard staristical graphics, and thc rccent invcntions of 

Chernof and Tukey. 

The observed increases in eliciency, in how much of the graphic's 

inL carrics information, are sometimes quite 1arge. ln several cases, 

the data-ink ratio increased fron,r or .z to nearly r.o. The transformed 

designs are 1es cluttered and can be shruak down rnore 

readily than the onghals. 

But, are the transforrned designs lener? 

(r) They arc nccesarily bcncr wirhnr the principlcs ofthc thcory, 

for morc information per unit ofspace and per unit ofink is displaycd. 

And this is significant; indeed, the lnstory of devices for 

communicating infonn:tion is written in terms ofincreases in 

eltrciency of communication and troduction. 

(z) Graphics are alnost always going to improvc as thcy go 

through edldng, revisior, and tcsting agaimt diFercnt dcsign options. 

Thc principlcs of ma-rinnznrg data-ink and erasing generate 

graphical alternatives and aho suggest a direction in whch revisions 

should move. 

(3) Then there is the audience: will those looking at the new 

designs be confused? Some of the designs are self-explanatory, as 

in the case ofthe rangeframe. The dot-dash-plot is more dilncult, 

although it sti11 shows all dre srandard infonnation fbund in thc 

scartcrplot. Nodring is lost to those puzzled by the fiame ofdashes, 

and sornething is gained by thosc who do undqstand. Moreover, 

ir is a lreqtent mistakc in thinking abour statistical graphics to 

underestinute rhc audience. tnstead, why not asume drat ifyou 

urderstand it, most other readers wil1, too? crrphics should be as 

inrcJJrgent ar'd soo|i.ri, "red r\ rl-e rccomp ) 

hg re\r. 

(+) Some ofthe new designs may appear odd, but dis is probably 

because we have not seen drem befbre. The conventional dcsigns 

for statistical graphics havc bccn viewcd thou$nds ofdmes by 

nearly cvcry rcadcr ofthis book; on thc other hand, the rangeframc, 

rhc dor-dasLplor, rhe white grid, the quarrile plot, the 

rugplot, and the half-facejust a few times. With use, dre new 

designs will come to look jusr as reasonable as the old.

Maximizing data ink (within reason) is bur a single dimension of 

a complex and multivariate design task. The principle helps conduct 

expcriments in graphical design. Some ofthose experiments 

will succeed. There remain, however, many other considerations 

in the design ofstatistical graphics-not only ofefrcienry 

but also ofcomplexity, structure, density, and even beauty. 

DATI-rNK MAXIMTZATTON 137

7 Mubifunctioning Graphical Elements 

The same ink should often serve more than one graphical purpose. 

A graphical elernent may carry data information and also perform 

a design finction usually left to non-data-ink. Or it might show 

several different pieces of d*a. Stuch nufuifunctioning graphical 

alernrnts, if designed with care and subtlety, can ef,ectively display 

compiex, rnultivariate data.r 

Consider, for e

14O TIIIORY O! DATA GRAPIIICS 

Building &ta measures out of the data increases the quantitative 

detail and dimensionality of a graphic. The stem-andJeaf plot 

constructs the ihsftibution of a variable with numben themselves: 

0l9 - 900 rccr 

h€ights of 218 volcan@s, lnn 100 fel. 

19 3 = 19,300 aer 

9476562 

9??19630 

69947 1 6654442t1 t009 I 50 

47665541209955t4?5 

99884433t92943336110? 

97 6666655 442221 gJ97 

0 

2 

1 

5 

7 3 | 

6 898665441077?61065 

1 988554311006521080?3 

8 6533t12t17 

9 37765542t9J0493 

l0 09u4131652t2 

ll 496Xmt63r 

t2 4542IL64 

l3 47810 

00 

t5 676 

l6 52 

t1 92 

l8 5 

l9 39?30 

The idea ofmaking every graphical clernort effective was behind 

the desi$ of the stem-andJeaf plot. In presenting his invention, 

John Tukey wrote: "If we are going to make a marL, it rnay as well 

be a meaningfrrl one. The simplest-and most usdul-meaningful 

mark is a digit."' 

Here. too. the &ta form the data measure. Note the bimodal 

distribution in the histogram of college studorts arranged by height. 

,"Some cnthic dd Senig.aphic Di5- 

!la}," in T. A. Bddoft,.d., Stdriri.al 

Papas in Hotul oJ AeotSe w. sftdecol 

(An6, Iow,, 1972), 

!. 296. 

Arim l-.Joiner, Living Histogrim,' 

Inttutioel Stdtitti.dl Pcww.4j G9? j), 

339-34o. But, for 6:rtha ilevelopments, 

s MuL F. Sclillitrg. Am E. WatLi$, 

ud Wilid WnkiB, h Htu]6 Height 

Bitu'l^l?" Thz A6di@ Statistkidi, s6 

(Aug6t 2@2), 223-2t9-

A distinguished graphic thar builds data rneasures out ofdata *,as 

designed by Colonel Leonard P. Ayres for his staristical hnrory of 

World WWar 

I, a book with several notable graphics al1 done by 

typewriter and rule. Consrructing rhe dara measues out ofeach 

American division's name (a numerical designation) tums wh:rt 

might have been a routine timc-scries into an elegant dnplay. (Note 

that rhe cumuladve design depends on the facr that none ofthe 

divisions retumcd before October 1918.) The triple-functioning 

data measure shows: (r) fie number ofdivnions in France for 

each rnonth, June r9r7 to October r9r8; (z) what particular divisions 

were in Fnnce in each morth; and (:) the duration ofeach 

division s prcsence in France. 

MUlTrtuNcTtoNINGGtApIrlCAtXrrMrNTS 141 

lit !a 

a 

& let 

a7 61 

.lo rlo 40 

t9 a9 5t 

a3 aa 3a 

al al 31 

? 7 7 

a5 35 0! 

15 !6 t6 t6 

79 79 79 f9 

fa 76 7a la 

'9,'2t29', 

a t q e t s q 

9o 90 90 90 90 

92 92 92 92 9a 

0, a9 a9 69 39 

6t 0t €a 6t 05 

?amtatat0 

a0 a0 !o a0 3o 00 

!o 50 !0 !o lo lo 

t! 5t 5:t t5 !! !t 

6 6 6 5 6 6 

tt zt zl zt dt 9l 

! t a a 4 a a 

2A2A?A,l2a?€ 

t5 55 t6 t5 66 !5 

a2 a2 62 A2 A2 AZ 

7t 7t 71 77 tt 7t t, 

! t 5 5 ! ! ! 5 

5 5 5 5 6 6 5 5 

az a2 a2 62 a2 A2 a2 A2 52 

al at ,l]' at 4t ro {!. ,11 .u ,11 ,tt 

&424.€.c42a2 {:42 1E 42 4,42 

26 26 A 26 26 21 26 e6 26 2A 26 6 

2 2 2 8 l a 2 2 A 2 2 2 2 2 2 2 

t l 1 l L 1 1 1 : . r 1 1 1 r l 1 

191? 191a 

Leomrd P. Al,res, ?r. t/,,,,id cer 

rJ,,r (wrslington,D.c., 

lere),!.io2.

142 TIIEORY OF DATA GRAPHICS 

Encoding of data measures can be frr more elaborate. The 

ploted points here are Chemofffaces, which reduce well, maintaining 

legibility even with in&vidual areas of.o5 square inches 

as shown.3 The analyst would observe the standard X-Y scatterplot 

and then turn to the within-scatter detail, seeking clusters of 

similar observations over the X-Y plane. Oudying faces and those 

inconsistent with others in the neighborhood-drey are, of course, 

strargers should be identified by observation number or name. 

o 

o o 

o 

,"n /al 

\-,/ \ -/ 

/_n 

1 '4, ,i6A 

rn L-_-] 

fPxA 

/:]\ 

, r t 

o o 

With cartoon faces and evm numben becoming data-measures, 

we would appear to have reached the limit of graphical economy 

ofpresentation, imagination, and, let ir be admined, eccentriciry. 

to Repreent Points in k-Dimtri(Dl 

sp e cnphic l\y," Jounal oJ the Anet 

im Stdtistirdl Alecidtion 68 (l@e 1e73), 

36r-168. For d apl,lication of Aces le 

cated over two dinmsions, sec Howard 

'Wainer 

dd David Thisfn, 

"craphic 

D^n tu lysi,s," Awadl Reri* oJ Psy- 

.nobsy, i2 (r98t, 19r-24r.

But let us consider this shaped poem, "Easter Wingi' by George 

Herbert (1593-1633), which uses space-the lorgth of each Lineto 

depict quantity, all done r5o years before Playfair.a The lines 

double-function: the longer Lines describe wealth, plenty, largese, 

and rising to flight; shorter Lines tell ofpoverry and becoming 

"most 

thinne"; and lines ofintcrmediate lenerh indicate rransirion 

:nd chrnge (decrying. ri(ing. combining. be;oming): 

-EASter-utn{s. 

fOrd, rrho createdst man in wealth and store, 

l_7 Though foolishly he lost thc same, 

Decaying more and more, 

Till he became 

Most Doore : 

wrth thee 

O let me rise 

As larks, harmoniousl6 

And sing this day thy victories: 

Then shall th; fall turihe.ihe ffight in mc. 

My tend€r age ih sorrow did beginne: 

And still with sicknesses ahd shame 

Thou didst so punish sinne, 

I nat I became 

Most thinne. 

With thee 

Let me combine 

And feel this daythyvicto c: 

For, if I imp mv'wiie on thine. 

Affiction shall a'doaice thi flight in-me. 

MULTIIUNCTIONING cRA?ErCAI, ELXMENTS 143 

a For a remrkable orsoc liketo!. of 

the maDy tFographicai varlant shapes 

of 'Esier Wingi in its long publcation 

hnrory, see lhe esay tr^r tLUx," by 

"Rrndom cloud" in Randall Mcleod, 

eC,.,C'8it tt Editing: Texx oJ the E'glish 

Rrfuisard (New Yo.k, 1994).61 172.

144 THTORY OF DATA GRA?HICS 

Fnrally, this graphical pun: the visual d:ta as the data measure, 

as in thc living histogram. The chart shows how states once dil 

{ircd in their engineering standards for painting lane stripes on 

road pavemenr. Some states marked the road lanes with short 

dadres and long gaps; others used only solid lines. Portrayed in 

the graphic is the acrual physical pattcrn painted on the road, with 

48 U.S. srates ordered by thc lcngth ofthe painted mark: 

Ohio 

1@ 1Jo 20o 25o 3oo 3Jo 

R€dnm Gom A. R- Laucr, Prycholosical 

Factos h Efec.ive Trafiic conttot 

Ddi.ccs, Ttdfrc Q,an$]L 5

Data-Based Grids 

Very occasionally the grid can repot dircccly on the data. This 

grid is formed by the locarion of measuemcnr instruments; the 

plain dots register a zero reading, in contrast with the white background 

where no readings were taken. Erasing the grid would 

erase measured data (rather unevcnt6rl, to be sure). Such is not 

the case for most grid dots, ticLs, and lines. 

The arrangement of data in this table-graphic yields an intemal 

grid, a rare exanple ofdara as grid: 

ADU!I CHILDREN 

r!e; lleiShE I ad !.viarioft sm 68!in ,lks, 

MUITIFUNCTTONING CRAPHICAL TTEMTNTS 145 

K. v. Robeirs md D. E. Porier, "MagnetohydrodyDnic 

Calolations," in 

Bemi Ald€r, et al-, edi., Merl,o6 ir 

Conpltational Phyti.t: Vol"ne 9, Pldma 

P/4/rid (New York, 1970), 

!. 4o2. 

Ktl Peasan, mo Life,lLfi,s mti Ldbout\ 

oI Ftun.i! Galtan lctmbideq

146 THSORY OF D,\TA GRA?IIICS 

_elndgh!TQvdM!Ds 

sJBiq,h nto! dd'4 rdsq. a 

b d! r,ir&r eiq vh{ Joi! Mrd 

vidddlhedigid)o9iidIi!qish| 

TL€ Unied Sab in Nodh Are.icl 

(Mirhell Mip) 

i.e +@is .h. &dsEd.d d & .sd Fd@ d Nod 

ic F4hc/ d Nod Aire 6d d. s 

liFole.ddei.lFid 

Here the grid is the elenent ofinterest, rather than the rrap. IrsEr J. Ca!!on, Barbara Batz Petche 

nik, md John Hamilton I@9, ,4116 o/ 

Eatly AneticM Hi$ory efir.etm, 196), 

P. J8.

The gnd that follows presents the data on rhe surface of the ro&; 

on the sides, the grid n conventional. The rwo displays comprre 

the efi'ect ofreligion, taLing into accormt parry afiliation, on a 

person's votc for preJdent in 19j6 and in 1960 (when a Catholic 

ran lbr president). Note there is no reliable slope associatcd widr 

religion in 1956, once party is controlled; in 1960, a sysrematic 

effect is found. Readnrg the slopes in the other direction shows the 

persistent efi'ect ofparty in both elecions: 

z 

g 

Weak Strong 

Independenl Republi€n 

E 

1@% 

75% 

50% 

25% 

w 

MULTIFUNCTIONING GRArrrrc4[ rrxMrNTs 147 

Plilip E. Convere, _Religion and Po1itics: 

The 196o Election. in Angus 

Canrlbell, Pltiip E. Convdsq Warrcn 

E, MiUe., andDoDld E. Stokcr, Eie.nonr 

ari ke Pohicdl On10 (tlew \ork, 

W*k Strong 

lndelend€nt Republican

148 THEORY OF DATA GRA?IIICS 

Playfair tled the grid ro the data in his skyrocketing debt graphic. 

Although thc implicit piotting coordinates are based on regular 

intervals, the verticalgrid lines nr the published version are irregularly 

spaccd, ke,r'ed to significant events. The data-based grid is 

a shrcwd graphicrl device, sewing rather than fighting with the 

data. It is a technique underused in contemporary graphical work.

Double.-Functioning Labels 

MUITIIUNClIONING 

Data-based coordinate lircs lead. to data-basei la6els, as, for example, 

at the boftom of?la),{air's debt graphic. Again, the issue is rhe 

same: why not use the ink to show data? Beginning with conventionally 

labeled frame 

0 

10 

and erasing to the range-frame 

_------.1 

2 0 3 0 { o 

leaves those lonely ticks and numbcrs out on the raiis, workine to 

help rhe eye ger a berrer readrng on nhere the line ofrhe ransF 

frame ends. Bur tharjob can be done better by inuestbg rhe irme 

ink in data: radrer than showing the minimum round number 

and the maximum round number at the ends of rire ftame, show 

the actual mininum and maximum realized in the data: 

10 

_ - - - - . ' - - - - 

20 30 37 

Witl itr greater precision ard rwo tick-marks less ofnon-datainL, 

the range-frame wirh range-labels is superior to rh€ range- 

Game with round number labels. Both improve on the shnalard, 

pasive frame. 

Numbers also double-function when used both to name things 

(like an identification number) and to refect an ordering. In this 

graphic (in which the circled numbers fail to double-function), 

each number idcntiies a parricular study ofthe rhermal conduc, 

tivity of tungsten, ordered alphabetically by the last name of the 

first author. Ifthat list were order-ed by &te ofpublication instead, 

then the code would also indicate the time order in which 

40 

GRA?EICAIEtttrlENTS r49

4.4 

3 

150 TIIIORY OF DATA CRAPI{ICS 

the various conduoivity determinations were made. Thus "1" 

would indicate the earliest study, and so on or, altematively, 

"6rc" 

would be the third study published in 1961. Such informadon 

has interest, since we could see which of the *rly studies got 

the right answer. tn addition, the movement ofthe studies toward 

the "conect" recomrnmded values could be tracked. This extra 

information requires no additional ink. 

THERMAL CONDUCTIVITY OF 

TUNGSTEN 

In most graphics, the coordinate labels are far from the dara 

measures. Consequcntly the eye ofthe viewer must move back 

and forth between the path formed by the data and rhe coordimte 

positions arayed along the margins ofthc graphic. Sometimes this 

eye-work c:n be eLiminated cntirely by turning the coordinate 

labels inro dara meesurcJ. Jnorhcr doublFftmcrioning maneuver. 

Take the exarnpie from the style sheer ofthe-/ormal o.f the Anetia 

Statisttual A:sociation: 

C-Y.Ho,R. W- Powell. dd P. E- Liley, 

Thdnal Cobiludivq a.f the Eknenrs: A 

Conpfthen\ite Rerie'!, srpplenle^t ô. 

r,lor al o.f Phy'nd| dnd Che\nat ReJierc 

Datd, j \r9J4),1492.

The grid inoement! ofthe X-axis are relocated upward to mark 

the path of the data: 

.15 

.10 

.05 

, t n ,o,r rr.ruros 

A:rd since the issue in this display is the probabiliry at each integer 

value, the round-number Y-scale is replaced by exacr values: 

.t77 

.rL4 

.u5 

.052 

.034 

.025 

.004 

.w2 

, , n TorTrr13ro$ 

The Y-scale now resembles rhe dashes ofthe dot-desh-plor. wirh 

rhe r ertical column of&u-positioned numben servine as tbe 

dasher to rndicete Lhe mareinal disnibudon. 

MUI,TIFUNCTIONING GRAPHICAL !lEMTNTS 1J1

1J2 THEORY OF DATA GRA?IIICS 

The method ofdata-based markers for the marginal distributions 

suggests a further enhancement ofthe dot-dash-plot: 

m.3 

L4.6 

11_3 

70.7 . 

8.4 

Now the numben in the margin eliminate the standerd frame 

and everr a rangeframe, replace the coordinate ticks, show thc 

marginal distribution ofboth variables, and rccord the exact values 

of rhe two measurements made on each udt of observation. Tlis 

graphical anangernent performs better for smaller data sets (say 3o 

observations or les) and when a fine level of detail is required. 

Fnrally, a srriking design with data-based labels: 

Desiened by Carol Moor€, Corpoate 

Amul Reports, Inc., in Wwaltd H-dcg, 

Gqhn/Diastdns (zwi&, r9j6), p. 23-

Puzzles and Hierarchy in Graphics 

The complexity ofmultifunctioning elements can sometrmes tum 

data graphics into visual puzzles, crypto-graphical mysreries for 

the viewer io decod€. A sure sign ofa puzzle is that rhe graphic 

must be interpreted rbrough a verbal rather than a visual proces. 

- For example. despite i* clever and multifuncioning dara measure. 

formed by crosing two four-color grids, this is a puzzle graphic. 

Deployed here, in a feat oftechnological virtuosity, are 16 shades 

ofcolor spread on 3,oJ6 counties, a monurnenr ro a soDhisticated 

computer graphics system.s Bur. it is surely : graphic eiperienced 

verbally, nor visu:Jly. Over and ovrr. rhe viewer' musr nrn 

linle phrases through their minds, trying to maintain the righr 

paitem of words to make sense out of the visual montese: i Now 

let's see, purple represents courties where rhere are both hieh leveh 

of nule cardiovascular disese mortality aad r r.6 to J6.o p;ent 

ofthe households have more trhan 1.or persons per room. 

. . . What does thac mean anl,way? . . . And the yellow-green 

counties. . . ." By conrrasr, in a non-puzzle graphic, the translation 

ofvisual to verbai is quickiy l€amed, automatic, and implicit 

- so th::r rhe rrsu.al image flows ight through the verbal decoder 

inirially necessary ro understand rhe grephic. As paul ''Seeing 

Valdry wrore, 

is lorgetring tfie name ofrhe tlunq one sees." 

MUI,TIFUN ClIO NIN G G&rlPrrrcar rLlMtNTs 1J3 

5The .r.chnique n desdibed in Vincent 

P. D&bb: ed Alva L. Iintner, The 

Utilizrion of Prim*y Printing ColoB 

in Displaying More rb,n One vdiatte," 

iD Buceu ofthe CeNui Technical paper 

No. 41, Ctuphitul Pr.htdtkh of Statistiral 

I"Iotadtior (wasbi\st n, D.i., 1s78), 

14-2r. The m1p! are as6ed in Howrd 

Wain€r aDd C. M. Francolini, ..An 

Empnid Inguny Concehing Hund 

UndeBbnding of TwGveriable Color 

M,ps, Anenan Sldtutittdn, 34 (1sqo), 

8143- 

!*:': 

! t"'' *"

1J4 THEORY OF DATA GRA?IIICS 

Color often gclerates graphical puzzles. Despite our experiences 

with the specrnrm in science textbooks and raiubows, the rnind's 

eye does nor readily give a visual ordering to colors, except possibly 

for red to refect higher levels than orher colors, as in dre 

hot spots ofthc cancer rnap. Attempts to give colors an order 

result in drose verbal decoders and thc mumbling oflittle mental 

phrase.s again-indeed, even rnnemonic phrases drotl the phras€s 

required for graphical decodiqg: 

A method ofcoloring ingenious in idea but aot very satisfactory 

in practice was used by L. L. Vauthier. It was called thc 

mountain-to-the-sea method. White was uscd for thc representation 

ofthe greatest intensity ofdre fact because it indicated 

the sunrmit of a mountain with its etemal snow, irext came 

greer reprcsenring the forcsts farther down tlc slopcs, thcn 

yelJow for the grain ofthe plains, and fnally for the ninimum 

the blue of the waters at sea level.6 

Becausc drey do havc a natural visual hiemrchy, varying shades of 

gray show varying quantities bcttcr than color. Ten gray shades 

worked ef'ectively in the galaxies map: 

The success ofgray compared to rhe visually more spectacular 

color gives us a lead on how multifunctioning graphical elemena 

can communicate complex information without tuming into puzzles. 

The shades ofgray provide an ea.sily comprehendcd order to 

the data measures. This is the key. Central ro mainraining clarity 

in the face of the complex are graphical methods that organize ond 

order the Jlow oJ graphical infornation presanted to the eye. 

How can graphical architecture promote the ordered, sequenced, 

hicrarchical flow ofinfotrnation from the graphic to the mindi 

eye? How can dre data-information be arranged so that the viewer 

is able ro peel back layer after layer ofdata from a graphic? 

Multiple layers ofinformation are cre*ed by nulupb t iewing 

deptlu and nubQle uiewing aryles. 

6H, Gay F@lhotrsr, Histoiiel Dt 

velopncnr of thc cnphn,l 

tion of Sbti*io1tub," OJ'tn, 3 ( 

126, who cites E. Cheyson, L6 D& 

thodes de statisnquc ghphiq@ il 

snion eiv6c11c dc 1878,"Jolb.ld.L 

Sadltl de *atnri|,e de Pdtn,I (9fi\

, Graphics can be desigftd ro have at le3st rhree viewing depths: 

(r) what is seen from a &stance, an overall sru.*.. *,rllly "ggr"_ 

gated fiom an lnderlyng microsftucture; (z) wh"t is "een op1t"e 

and in detail, the frne structure ofthe data; aad (3) what is seen 

1rulicidr,.gd1lylng 

ttre graphic- that which i, U.ti"a th. gophi". 

Look at all the difi'erenr levels ofdetail created by thts popj"ion 

densiry map ofLhe Unired Stares, a glory ot -"a.r" ."rog.rphf 

prepared by rhe Bureau ofthe CersLrs rr.l, dot. .*..pr tn"url"o 

centers, represe[ts jcrc peoplel Note the corridors connecting the 

major urban complexes; the e6ec* ofJandforms on th" popJl"tion 

di"smbudon lrhe cennal valJey ofCrlifornia. the raJlrys'and ridges 

oI Appahchra. and the clu,rer. along ;ven1r ana the imalt rown'. 

erong rhe highways, linled like; srring ofpearls. The map arrays. 

Inrde-cr. lome 4oo,ooo points on its implicir grid. 

Differenr visualangles for differenr aspecrs ofrhe dau aLo or_ 

ganize graphical inlorm:rion. f,ach separare line ofsrght sbould 

remain undranging (preGrably horizontal or verricel) as the eve 

watches for dara variation of the fat ofthe Iine ofsisht. For mulrivariare 

wor[. sereral clear lines can be crerred. nccJJ Ayres. 

display ofAmerican divisionr in France. Even with its comolex. 

inrerwo\en dr(a.th€ grrphic is nor a puzzJe. Tt ... .eg"rar. ;.*l 

angles make the flow ofinformation coherent: the profle of the 

horizon for rhe upward-moving time-series, the vertical for the 

composition ofthe bar, and the horizontei for each division's stay. 

Thus while every drop ofink serves three dillerent data dispiay ' 

functions, each of the tluee comes to the eye v/ith irr ovrn inde_ 

pendence and integrity. 

..,iiii+iE{ttitti 

--.**ffi-* 

MUITIFUNCTIONING GR4PIIICAL 3!EMSNTS 1

158 TTTEORY O! DArrl GRAllIIcs 

Current R€ceipts ofcovemment as a 

Percenage of Gros Domestic 

Prcd\\cr, rg7o ^nd. lgjg 

Belgiu 

1970 1979 

37.5 

34,9 

Ir.ly 3o.4 

43.2 Sdgim 

3J.? I$ly 

27.1 Sp.in

, SimiJarly. this rable-gnphic organize dau for viewing in sevenl 

drrecions. Tbe chart, wber read vertically. r.arr[, ,5 ."un;es by 

govemmmt tax collections in r97o and again in ry79, with the 

names spaced in proportion ro dre percenrages. Acros *e mlurrurs, 

rhe parred comparisons show how rhe numbers chaneed over rhe 

yean. The dopes are also compared by readirg d"*n ,t " - .._ 

tjon oflines. and lines ofunuruai slope srand our 6om rle overalt 

upward panern. The informarion pftown is boLh integnted and 

separated: integrated chrough im donnecrJ conrrnr, separared iD 

that tLe eye follows several difiLlent and uncluttered patlu in look_ 

ing over the &ta: : 

Such an analysis ofthe uieuhg archttr.tnrc of a gr:Lphic will help 

tn aDd 

-creating 

evrluadng designs Lhar organize complex 

inlormadon hierarchically. 

MULTITUNCTIONING GIAPIIIC LEIEMSNTS 1J9

I uant to rcach that state of tondansatioh 

oJ scl$ations 

uhith .onstitutes a pi.tute.

8 Data Density and Small Multiples 

Our eyes can make a remarkable number ofdistinctions within a 

small area. With the usc ofvery light grid lines, it is easy to 

locate 625 points in one square inch or, equivalently, too points 

in one scluare cenrimeter, 

Or consider how an 8o by 8o grid over a squarc inch-about 3o 

by 3o over a square centimeter-divides the space:1 

With the help ofconsiderable redundancy and conrexr, our eyes 

nake fine disrinctions ofthis sorr all the time. Measuremenr insrruments 

used in engineering, architecrural, and machine work are 

engraved with scales ofuo increments to thc cenrimeter and 50 

to the inch. Or consider the reading ofline prinr. The rype in rhe 

U.S. Statistual Abstrca is set at rz lines per vertical inch, with each 

line running at abour z3 characers per inch for a manmum density 

ofz76 characters per square inch. The actual density, given 

the white space, is in this case r85 characters per square inch or 

28 per square centimeter. 

No, 1,t50, SrE L PRoDUCB-NE1 SsrlMEMs, By MAnnr CussEs: 1960 ro 19?8 

.. " na4 lot eb.vhse dNjnedl 

co.dsbdi]eebgoB.$'|pl'iP'.. 

T{ 

l1g 

1 A squre grid formed on c,c! side by 

! pdald black,nd n 1 pmlel vlite 

lncs contains D2 iniersections of two 

black lines (coners ofsqlrr€r, (n 1), 

intenections oat*o white liDes (whire 

sqEret, dd,n(n 1) ineseciions ofl 

black ard whnc [n. (sides ofsquder, 

for. total of (2n-r),llne inteBediols 

U,S. Bueau of the Ce$us, Srd,s/kdl 

Ab4tdtt o.f rhe unit.d stdtes: 1979 (tv.shingron, 

D.C., 1979), p.822.

1O2 THIORY OI DATA GRA?IIICS 

Maps routinely prescnt even finer detail. A cartographer writes 

that "the resolving power of the eye enabler it to diflerentiate to 

o.r mrn where provokcd to do so. Clearly, $erefore, conciseness 

i.s of the essence and high resolution graphics are a common 

denominator of cartography."2 Distinctions at o.1 inm mean 

254 per inch. 

How many statistical graphics takc advantagc ofthe ability ofthc 

eye to detect large amoutrts ofinformation in small spaccs? And 

how much informatioa should graphics show? Let us begin by 

considering an empirical measurc ofgraphical pcrformancc, thc 

data density. 

Data Det$ity in craphical Practice 

The numbers drat go into a graphic can be organized into a data 

maftix ofobseffrtions by variables. TaLing into account the size 

of the graphic in relrtion to the arnormt of data displayed yields 

rhe data densitll 

data density of a graplnc : 

number ofentnes rn dau m:mx 

area ofdata graphic 

Data matrices and data demities vary enormor:sly in practice. 

At one extreme, this overwrought display (originally printed in 

five colors) presents a data natrix offour entries, the names and 

dre numbers for the two bars on the righr. The left bar is merely 

the total ofthe other two. The graph coven 20.5 square inches 

(r7r squarc ccntimeters), 

resulting in a data dcnsity of.r5 nurnbers 

per square inch (.oz numbers per square centimeter), which 

is thin indeed. 

,D. P. Bi.knore, 

"The 

Relmc of 

cartogmphy," inJohn C. Davis md 

MichaelJ. Mcculash, eds., Dirpial dfld 

,4nalysn oI Spdtidl Datd (L@dot, rgj s), 

P 331

TOTAL 

PARTICIPATION 

DATA D!NSITY AND SMAIL MULTI?TIS 103 

lnadult 

-35 

30 

25 

z0 

Exc@dve Onice ofthe Prcsid.nt, Olice 

of MMgcftent and Bldget, Sd.td1 

Ldi.at r, rr7_r Nqrshingron, D.C.,

164 THEoRY oF DATA GRAPIIICS 

The exernplar ftom theJASA style sheet comes in at a lightweight 

3.8 numben per square inch (o.6 numbers per square centimeter) 

and a small data natrix of 32 enrries: 

ln contrast, the New York wearhet history, in this reduced 

version, does very weii at r8r numbers per square inch (28 per 

square centimeter):

An annual sunshine 

rccord reporrs about r,ooo numben per 

square nrch (16o per squarc cenrimcter): 

Thc visual nctaphor corresponds appropriarely ro the data ifthe 

image is reverscd, so that the light areas are the tines whcn the 

DATl Df\5I-! A\D SI1A.t vtttI?l!S rh< 

F.J. Mo*house and H. R. WitkinsoD, 

M4," dla DidCldr4 (lond.n, rhnd 

ednion, 1971), 

!!. 2+L2,13.

166 THxoRY or DATA GRA?Hrcs 

This map (:7 square inches, r75 square centimeters) shows che 

location and boundaries of 3o,ooo communes of France. It would 

rcqlrire at least 24o,ooo numbcrs to recreate the data ofthe map 

fuo,ooo latitudes, 3o,ooo longitudes, and perhaps six numbers 

describing the shape ofeach commune). Thus thar data densiry 

is nearly 9,ooo numbers per square inch, or r,4oo numbers per 

square cenqmcrcr, 

The new map ofthe galaxics locates 2,27J,328 cncoded recrangles 

on a two-dimensional surface of61 square nrches (39o square 

centimerers). Each rectaDgle represents thrce numbers (two by ia 

location, one by its shading), yielding a data density ofrro,ooo 

numben per .qu.rre incf or 1-.ooo numbers pe( rqurre cenrimerer, 

That is the current record. 

J,.$es Bertiq Se,tol,rie Crdprt ,e 

(?aris, sc.ond €didoq, 1e7j), !. 1j2.

Data Density atld the Size ofthe Data Matrix: 

Publication Pnctices 

DATA DlNSTTY AND 

The able shows the data density and the size ofthe data matrix 

for graphics sampled from scietrrfic and news publications. At least 

zo graphics from each publication were examined. 

Thc table records an enonnous divenity ofgraphical performances 

both within and betwecn publications. A fcw data-rich designs 

appear in nearly every publication. Thc opportunity is there but 

it is rarely exploited: the average published graphic is rather thin, 

Data Dasity and Size df,Data Mitrix, 

Statistiel cnphics in Sel€td Publicatiols, Cn@ 1q79 1980 

oruben !d squre in.h) 

medio rnnimum maxnrq 

n 

2L 

19 

18 

18 

t7 

13 

t2 

9 

8 

8 

7 

7 

6 

5 

5 

2 

o.2 

3 

5 

f 

5 

2 

2 

3 

1 

I 

1 

1 

1 

2 

1 

7 

1 

2 

362 

115 

754 

1l 

122 

lo 

113 

923 

5L 

17 

74 

39 

1X 

72 

t3 

25 

69 

23 

10 

Sir ofData MaEix 

m€did hinimum ruihm 

r77 

2ctr 

lo9 

0.1 1 

5 

2!J 

% 

50 

150 

29 

84 

36 

35 

32 

23 

44 

3a 

16 

15 3780 

10 1,160 

376 

2A 1aA 

42 156 

14 44A 

46 I&fi 

L5 472 

I 3600 

3 192 

1L 3t2 

a 420 

14 735 

6 

t4 

2 

t2 

580 

96 

96 

540 

14 652 

I 164 

sMAl! MULTTPL!S 107

168 TEEoRY o! DATA CRA?IIICS 

based on about 5o numbers shown at the rate oflo per square 

inch. Among the world s news?apen, the Wall Steet Jounal, The 

T;nes (rondon), and zrali publish data-rich graphics, with data 

densities equal to those o{ the Jowul of the Aneican Statistical 

,4rsoriarron. Most ofthe Amencan papers and magazines, along 

with Praala, publish less data per graphic than the major papers 

of other industrialized countries. 

Very few statistical graphics achieve the information display 

rates found in maps. HiglJy detailed naps portray loo,ooo to 

lSo,ooo birs per squarr inch. For example. che average U.S. 

Geological Survey topographic qua&angle (meauring r7 by z3 

inches) is estimated to contain over 1oo million bits ofinformation, 

or about z5o,ooo pet square inch (4o,ooo p.r squ"r" 

centimeter).3 Perhaps some day staristical graphics will perform 

as succrssfLill) as mJps in carrling infonnation. 

Iligh-Information Graphics 

Data graphics should often be based on large rather than small 

data matrices and have a high rather than low data density. More 

information is better than less information, especially when the 

marginal costs ofhandling rnd interpreting ad&tional info nation 

are low, as they are for most graphics. The simple things belong 

in tables or in the text; graphics can give a sense oflarge and 

complex data sets that calrnot be managed in any other rnay. If 

the graphic becomes overcrowded (although several thousand 

numbers lepresented may bejust fine), a variety of&ta-reduction 

techniquet averaging, clustering, smoothing-can thin the numbers 

out before plotting.a Summary graphics can emerge from 

high-information displays, but there is nowhere to go if we begin 

with a low-information daign. 

Data-rich designs give a context and credibility to statistical 

evidence. Low-informstion designs are suspect: what is left out, 

what is hidden, why are we shown so little? High-density graphics 

help us to compare parts ofthe data by displaying much information 

within the view of the eye: we look at one pege at a time and the 

more on rhe p€e, de more efective and comparative our eye 

can be.s The principle, then, is: 

Maximize &ta density and the size ofthe data 

matrix, within reason, 

High-information graphics must be designed witb special care. As 

the volume ofdata increases, data measules must slrink (smaller 

dots for scatters, thinner lines for busy time-series). The clutter of 

I Mo[is M. Thompson, Ma?Jrr ,u ertd 

(washinston. D.C.,1e7e). p. 187. 

'Paul 

A. TGey sdjohn W- Tukey, 

''s@mriatior: 

Smoothing: Sulplemenied 

Views," in Vic Baben, ed., 

Iiteryrcting Mdnuatidt, Ddta (Chid\6t{, 

lDgland, 1e32), ch. 12r dd william s_ 

Cleveland- "Robst Loely wcighrcd 

Resre$ionand SmoothinsScatterplots," 

Joufldt oJ &e Anqrrdn Sratkti.dl AMtiation,74 

(1979J, E?D 336. 

slt is sucgstcd in the am\ris ofx{ay 

frlms to "smrch a reduced imse so that 

the whole dirllay cm be pereived on 

.t lest one o.6ion without large eye 

novement." xdwud LlewcllF 

Thoms, "Advic 

to the Ssiclld or 

'What 

Do We TeU Tbem?" in Richrd 

A. Monty andjohn W. S@dds, eds., 

Eye Moteftentt tud Pstclblogkal Ptur.ss! 

(HiUsd.re, N.J., 1926), l. 349.

charqjunk, non-data-ink, and redurdant data_ink is cven more 

costly than usual in data-rich daigns. 

The way to increase data derxity other than by arlarging the data 

matrix is,to lcduce the arca ofa graphic. The Shrink pn;ciple has 

Graphics can bc shruaL way down. 

Marry data gmphics can be reduced in erea to halftheir currently 

publrshcd nze uith rirru:ll1 no lo- rrr legrbilirl and rnforrnadon. 

for e),ampje. Ber rin s crisp ,nd eleganr line rllow. the jsplal 6f 

r7 "nrll-scJe graphics on a single pagc aJong wiLlr exren.irerexr. 

Repcared.applic,rion of rhe Shrinl princ pte lead-r ro a pou erlii 

and effecrir e gr:phicrl desrgn. the small muJnple. 

^tr--: 

ll_] 

ihfilifltl'ml] 

lMl' 

i 4 - - 

&YW+\l 

'E= 

i' 

-- 

ilruulllil 

Fffi 

.l'-l 

) 

lL*./ I 

t3T TI rl 

t\h4'!h1 

Lrl 1l 

lHl' 

DATA DINSITY AND SMALL MUTTI?TES 169 

POSES PAR LES CHRONIOUES 

;lt:',i:ip.tT,"t:;l:ii":.* ^',''" p,. d.* 

co!rb

1'O TH!ORY O! D-{TA GRAPHICS 

Small Multiples 

Snall nruldplcs rcscniblc the lrames of a nrvic: a scries ofgraphics, 

shosng the samc conibrnaron ofvariablcs, n,dcxcd b.v chrnges 

in aro cr vanable. Trvenq.-tlucc hours ofLos Angclcs air pollution 

arc organized into drn drsp1a,v, bascd on a cornputcr gcneratcd 

vidco tape. Shown is the hourlv :rvcragc distribution of 

reactiic h,vdrocarbon emisions. The design rema,Ds corlstant 

tlrrough all tLc f.arnes, so drat ancution x devotcd cnrirclv to 

slrrfts in the data: 

I'.on vid.. trte b-! Grego.\ J Nl.R:. 

Crlilon!r Iin.nne oaTdhn.l.g!. 

Th. n,od.l n deribed in c.J. I4cRr. 

Ww. R. a;oodnr, andJ, H. S.nrf.]d. 

''DevelopDrent 

of tr Sc.ond-Cicncrro.r: 

Nhdr.Dr.i..li!4.dci Lr UrbrL An 

!ollun... L Mod.l F.mulxtior,' 

Ah"oijl)nr Lrrnan"Ert, 16 (198)),

These grim small multiples show the distribution ofoccurenc€ 

ofthe cancer melanoma. The sites ofz69 primary melanomes are 

recorded, along with the distribution between men and riromen. 

Note the data graphical arithmetic, similar to that of rhe 

multiwindow plot. 

Abb. 1. Yerteilug von t69 pnmdm l[eleodm rnj Kopf 

md Ea,ls 

Abb. 2 

Abl. 3 

Abb,2 u,3. Diff6@ierung dor MeleDon@t€itunq 

@h G@NtuLt€n 

DATA DINSITI AND SMAfL MULTIPIES 171 

Arihu Wistelm, 

"Zu 

Mddom€ntrtehug 

dtrch chronishe Li.hreinwiF 

k@e," Dd Eautaftt,2s (rs71), t1.

172 THXORY OI DATA GRA?IIICS 

The efi'ecs ofsampling errors are shown in these rz disuibutions, 

each based on a sample of 5o random normal deviates: 

j"1 

/.I./rJL-.-r 

*:] 

'l ,J-\1,r. Iaa. 

Thcsc six distributions show the age composioon ofherring catches 

each year from r9o8 to r9r3. A ttemcndous numbcr ofhcrring 

were spawned in r9o4, and that clas began to dominate the 1908 

catch as four-year-o1ds, then the rsos catch as {ive-year-olds, and 

at5676q$11 12 I11 1516 

1908, 

1909. 

h 1910. 

r+L 191'1. 

) l\\ 1e12 

/ \ re13, 

5t56iE9m1n6h15t6 

This next design compares r comglex set ofdata: shown are the 

chromosomes of (from left to right) man, chimpanzee, gorilla, 

and orangutan. The similarities between humans and the great 

apes are to be noted. 

xdmo'dA. Muphy, "One caqe? 

Many Catres? Thc Argumflt from the 

Binodal Ditiburion," Jo,',al oJ 

chtonit Dives,17 (re6d, 3ae. 

Johan Hjori. "Fluctlario$ 

in rle Grc,t 

Fishoies ofNonhem E!rope," Rdr'pdt 

et Ptoteev,th4s,, 20 (1e14, tn Stsn 

S.blee, The Edge oJ e UnfMiliat Watld 

(New Yorl, 19?t, !.226. 

Jorse J. Yuis dd Om Phkash, "The 

O.jgjn of Mdr ,{ Cnromoionrr Pntorial 

Legacy. ' srrfte, 21j (March 19,

:in 

nEn 

|':t-' 

trt 

lt! 

!tFftl 

rlt 

tii 

trt 

!!" 

IFi 

l.ts l 

'l!l t - - 

ltt 

t,,1 r I 

l l l 

-a:1 

83tr 

iE- 

III 

=F! 

iti 

=FF 

,TEB 

TII 

!." I 

. t - l . l 

l t t 

-rit|: 

i F r 

I F E 

:!lli5 

! i t 

E:: 

! t l 

F ; : 

t t t 

i i t 

.1 

!!: ti 

l l t 

."it l l 

III 

r-i ul 

1a 

II 

T 

l.'l 

! 

: 

i:t 

r i 

t! 

tJr 

n': 

FI 

tt 

-fl 

!l 

0.' 

ili 

ilr 

FEF 

! i ! 

,''r 

! ! 

il 

rll 

!!! 

19 

Li!t 

I t t 

tlt- 

i:E 

8Fr I 

tt! 

F: :t t: 

llt 

-r ! t_l 

!:: 

itt 

n:a 

niir 

!q r r.l 

l t l 

:1 r"r 

-Fi 

!!! 

trrl 

!r'fl 

20 

...1 

!!I 

il! 

l r'! 

!ti 

I t t 

! t i 

l t - 

=tt 

fl -, ll 

ttt 

ltt 

ilI 

tft 

l r t 

iti'l 

lti f 

itt 

!r! 

t-t l-t rl 

!!: 

TTI 

Itt 

!tl 

lrl 

'l 

l' :l 

E Et 

|lf- l-i -.t -.1 

:i- r,ii:l E=:= 

ii: ;iii! .i::l 

iE! ',: ! ii +. i :: 

ilr I!-:= l==:= 

10 

t.lLi!: 

Frl 

HT 

,ir'! lir 

i iii 

--ii 

16 

!t! 

2l 

,E: 

11 

i 

f ! i 

itE 

t = i 

!3t 

!-!l 

'17 

12 

tiE 

trf 

FE: 

rll 

l l t 

llr 

ttr 

F-F 

lrt 

iLlF - II 

':-'rri 

rl ! !:

174 TIIEORY OF DITA G{APEICS 

And, faally, a visually similar small multiple, the Con*ner 

Reports fteqnenctafaepair records for autonobiles built from 

1926 to 1981. This is a panicularly ingenious mix oftable and 

graphic, portralng a complex set ofcomparisons berween manufacturers, 

tJpes ofcan, yeer, and trouble spo6. 

o-ed*tu-4' o=|.futuddn o-r@ o-$!{tuao( a=tu*'ftil'ft, 

ooooo 

ooooo 

ooooo 

ooooo 

ooo oo 

ooooo 

ooooo 

ooooo 

ooooo 

ooooo 

ooooo 

oooo 

oooo 

oooo 

oooo 

aoaao 

ooooo 

ooooo 

ooooo 

oooo 

oaooo 

ooooo 

ooooo 

oooo 

ooooo 

ooooo 

oaooo 

ooooo 

ooooo 

ooooo 

ooooo 

ooooo 

ooooo 

aoooo 

ooooo 

oooo 

oaooo 

ooooo 

ooooo 

oooo 

oo ooo 

ooooo 

ooooo 

oooooo 

oooooo 

oooooo 

oooooo 

oooo 

ooooo 

aoooo 

oooo 

oooo 

oooo 

oooo 

oo 

oooo 

oooo 

oooo 

oooo 

oooo 

ooao 

ooao 

oooo 

oo 

o oooo 

oooooo 

oooooo 

oooooo 

oooooo 

oo oooo 

oooooo 

oooooo 

oooooo 

oooooo 

oooooo 

ooooo 

ooooo 

oooooo 

oooooo 

oooo 

oooo 

CoMni Retfit, a7 ( ptn1982), 

oooooo 

ooooo 

aaoooo 

oooooo 

oooooo 

oooao 

ooo oo 

ooooo 

ooo ooo 

ooaoo 

oaooo 

ooaooo 

oooooc 

oooooo 

oo ooo o 

ooooo 

oooao 

ooooo 

ooooo 

ooooo 

ooooo 

oooooo 

oooooo 

ooooo 

ooooo 

ooooo 

ooooo 

aaooo 

o oooo 

o oooo 

ooo oo 

oooo 

oooo

Conclusion 

Well-designed snull multiples ate 

. inevitably comparative 

. deftly multivariate 

. shruaken, highJarsity gaghics 

. usually based on a large d.L -"trii 

. drawu almost entirely with daa-in_k 

' often narrative in content, showing shifts in the relationship 

between variables as the index variable dranges (thereby 

revealing interaction or multiplicative effectsJ. 

Small multiples reflect much ofthe theory ofdata graphics: 

For non-data-ink, le*s is more. 

For data-iok, less 

is a borc.6 

DA:lA DENSITY AND SMALI, MUI.TIPLES 175 

6The iwo alho.isms o the nani-c of 

'%s" are, readivcly, cdited to iu& 

wis Mi6 Ie dd Rohc dd io Rohst 

V&tnti, Convbitf 6rt Corbedkne i^ 

-4rdrtt 6.r (Nry YdL, s@d.diti@, 

tw),p.a?.

s""'"i. r r " r"' 

eark ,frauraliua a,, o.t"r^*'-L.-. 

" 

*.t:^- + a",al,J â-* a 6w. 

6\,L 

^ Ar. ^ h"s,*-'14 

r :x-t Li6d,4q&+ 

Ct"J. | 'r- att ) 

'q*-^ded@ 

-La'1""t* 

c4trai 

ug 

?Rlr s,.ltttr! 

&rtz &a"nlu * *"*-.8--- rr'o^,t Ja"a. tuL d@ ' .X@n ^stz -t8ts. 

" 

tu)+.^ 4t a) La--q-t ,.t ",4;a :.'^* q-- Z u,a--n ,*" 

-t:t*;EE:*fr . t *':;:-H.::*,: ; t t x;3 

;q1t .atr. rLqa, Lqv. L.fae Lc.-trls -4r +4 J.rt./;* "j'* + L 4 aLL; 

1..-;Fan:uat.'*rn;'l;>*rlLaL*;z=::x:8.*"*^^ 

dd&i AR;e!&d:!a 

4Lr.-,b?;1e-;";.t"F.- 

'.t ',tL*.!t-b'*\ 

;-' 

- "i29,=ffi=t*t 

.'

9 

t . 1 . l a 1 ' 

Aesthetics and Technique in Data Graphical Design 

Along with the anazing graphic ofthe French losses in the Russian 

invasion, Minard includes a second 

"Carte Figurative." Ir portrays 

Hannibal s fading eiephant campaign in Spain, Gaul, and Northern 

Italy. Minad uses a light transparent color for flowJines, allowing 

the underlying type to show through. Thh relined use of color to 

depict more information conrrasts with the garish tones too often 

seen in modern graphics. 

Whar makes for such graphical elegance? What accounts for 

the quality of Minard's graphics. ofthose of Playfair and Marey, 

and of some recent work, such as the new view ofthe gala-xies? 

Good design has cwo key elements: 

Graphical elegance is often found in simplicity 

ofdesign and complexiqr ofdata. 

Visually attractive graphics also gather power from contcnt and 

interpretations 

beyond the immediate display of some numbers. 

The best graphics are about the usetul and imporrant, about life 

and death, about the universe. Beautiful graphics do not tra{fic 

with the trivial. 

On rare occasions graphical archirecture combines rrr'irh th€ dara 

content to yield a uniquely spectacular graphic. Such performances 

can be described and rdmired but there are no easy compositional 

principles on how to create that one wonderful graphic in mi11ions. 

As Bamett Newman once said. "Aestherics 

is for the afist Like 

ornithology is for the birds." 

What can be suggested, rhough, are some guides for enhancing 

the visual quality ofroutine, workaday designs. Attractive displays 

of statistical infomarion 

. have a properly chosen format and desig. 

. tse words, numbers. and drawing together 

. teflect a balance, 

a proporrion, a sense 

ofrelevant scale 

. display an accesible compleity of detail 

. often have a narrative quality, a story to rell abour rhe data 

. are drawn in a profesional nanner, with rhe rechnical details 

ofproduction done with care 

. avoid content-free decomtion, including chafijunk. 

Cnades Joseth Minard, Tdrler,, c'e 

thiqu* et Catta Fiqnarius de M. Minahl, 

'r4J ir69, a pofifolio ofhis work held 

by tle Bibliothiquc de l'Ecole Nationale 

de Ponts et Chau$des, Paris.

178 TIIEORY OF DATA CRAPSICS 

The Choice of Design: Sentences, Text-Tables, Tables, 

Semi-Graphics, and Graphics 

The substentive content, extmsiveness oflabels, and volume and 

ordering ofdata all help determine the choice ofmethod for the 

display of quantitative materials. The basic structr.ues for showing 

data are the sentence, the table, and the graphic. Often two or 

three of these devices should be combined. 

The conventional sentorce is a poor way to show more than 

two numbers because it prevmts comparisons within the &ta. 

The linearly organized flow of words, folded over at arbinary 

points (decided not by content but by the happenstance ofcolumir 

width), otrers less than one efi'ective dimension for organizing the 

data. Instead of: 

Nearly 53 percent of the type A 

group did something or other 

compared to 46 percent ofB and 

dightlymore than 5 7 percent ofC. 

Arrange the typ€ to facilitate comparisons, as in tlns text-rablei 

The three groups dift'ered in how 

they did something or other: 

GrouP A 53% 

GroDpB 46qo 

CnonP C S7o/o 

There are nearly always beaer sequences *lrr alphabetical-for 

example, ordering by content or by data values: 

Group B +6Vo 

Group A szEo 

Group c stVo 

Tables are clearly the best way to show exact numerical values, 

although the entries can also be arranged in semi-graphical form. 

Tables are preferable to graphics for many small data sets.1 A 

table is nearly always better than a dumb pie chart; the only 

wone design than a pie chart is several ofthem, for then the 

viewer is asked to comparc quantities located in spatial disartay 

both within and between pies, as in this heavily encoded example 

ftom an adas. Given their low daa-density and failure to order 

numben along a visual dimension, pie charrs should never be used., 

.*-4 

Depanhdt of Suseys, Ministly of 

hlour,.4d6 d/Ivdel 0roten, 

19J6 ), vol. 8, p. 8. 

1 OD the deden of abl€s, @ A.S.C. 

"Rudimots 

Ehr6bdg. 

of Nmmcy," 

Joutul of the Roydl Stunakal Socieq, 

A, 14o (!97i,277-297. 

,t 

'This loiDt is n.de decisively inJacqGs 

Ber.n, Aryhia dnn Qa ic Infohati@ 

r@ari'J (3dlin, 1981). Benin dgcribes 

multiple pie charts s 'tompletely

Tables also work well when the data 

Presentation requires many localized 

compadsonl In this 41o-number table that I 

designed for the Neu Yol& Tifl$ to show how 

different people voted in presidential elections 

in the United Sates, comperisons berweefl rhe 

elections of 1980 and 1976 are read across each 

line; within-election analysis is'conducted by 

reading downward in the clusters ofthqee to 

sevm lines. The horizontal rules divide the data 

into topical paragraphs; the ows are ordered so 

as to tell an ordered story about the elections. 

This type of elaborate table, a supertable, is1*ely 

to attract end intrigue readers through its 

organized, sequential detail and reference-liLe 

quality. One supertable is far better than a 

hun&ed iitde bar charts. 

Ne yor& Tirrr November 9, 1980, 

!,-^ 28. 

ArsTrrBTrcs aND tEcHNtQuE 179 

How Diflerent Groups Voted for Prestdent

18o TFEoRY oF DATA GRA?Hrcs 

For sets ofhighly labclcd numbers, a wordy data graphic 

coming close to straight text-works well. This table ofnumbers 

is niceJy organized into a graphic: 

Matring Complexity Accessible: Combining rVords, 

Nunbers, and Pictures 

Explanations that give access to the richness of the data make 

graphics more attractive to thc viewer. Words and pictures are 

sometimes jurisdictional enemies, as artists feud wirh wirers for 

scarcc space. Arr unfortunate legacy ofthese craft-union difi'erences 

is the artficial separation ofwords and pictures; a few sryle sheers 

everr forbid printing on graphics. What has gonc wrong is drat the 

techniques of production instead of the information conveyed 

have been given precedence. 

Words and picturcs belong together. Viewers need the help that 

words can provide. Words on graphics are data-ink, naLing 

eft'ective use of thc space &eed up by erasing redundant and nondata-ink. 

It is nearly always helpful to write litde messages on the 

ploning iield to explain the &ta, to label oudiers and itrteresting 

dara points, to write equations and sometimes tables on ghe graphic 

itself, and to integrate the caption and legend itrto the design so 

that the eyc is not required to dart bacl and forth beoveen rextual 

material and the graphic. (The size ofqpe on and around graphics 

Netu Yo* Tin6,J^nwy 2,1979,p. D 1.

can be quite small, since the phnses and senterces are uiually nor 

too long and therefore the small tne will nor fatigue viewers 

the way it does in lengthy texts.) 

The pturcipb of data ltext integatiol is 

Data graphics are paragraphs about data and 

should be treated as such. 

Words, graphics, and tables ire dif'erent mechanisrns with bur a 

singlc purpose the presentation ofinformation. Why should the 

flow ofinformation be broLen up into dift'erenr places on the page 

because the information is pacl:ged one way or another? Somerimes 

ir may be useful to have multiple story-lines or multiple 

levels ofpresentation, but rhar should be a deliberate designjudgment, 

not something decided by conventional production requirements. 

Imagine if graphics were replaced by paragraphs of words 

and those paragrapls scattered ovcr rLe pages out of sequefte with 

the rest of the text-drat is how graphical and tabular information 

is now treatcd in the layout of many published pages, parrict arly 

in srientjfi c journrlurd profesronal book.. 

Tables and graphics should be run inro rhe texi whenever possible, 

rvoiding the clunxy and diverting segregarion of "See Fig. 

2," (figures all too often located on the back ofthe adjacent page).3 

Ifa display is discused in various parr of the text, it might well 

be printed afrcsh near each reference to it, perhaps in reduced 

size in larer showings. The principle of te*/gnphic/table integration 

also suggess th:t the same typeface be used for text and 

graphic and, further, that ruled lines sepanting diferent types of 

information be avoided. Albcrt Bidermaa notes that illustratiorx 

were once well-integrated wirh rext in scientifc manuscripts, such 

as those ofNewron and Leonardo da Vinci, but that statistical 

graphics became segregated from te* and table as prinring technology 

developed: 

The evolution ofgraphic methods as an elemenr ofthe scientific 

enterprise has been handicapped bv therr adiuncive. seeregated. 

:nd margrnal po.ition. the'exigencrJs of rypog.lphy 

rhar_moved graphics ro a segregared position in rhe prirred 

wo'k ha\ e in rhe past conrribuled ro rheir inrellecruJ segregrnon 

and marginaliry as wcll. There wx a correspondinq 

orgrnizarional segregatJon, uirh decisions on graphic' often 

pasrng out ofrhe hands ofthe orisinal allsi anid communicaror 

Into tho\e of qraphic .pecjalisr-the comrnercial arri*, 

and de'igncn of eraphic departrrenrs and audjo-vjsual aids 

shops. fiir example, whosc predilecrions and st

182 TExoRY oF DATA GRAlItIcs 

Page after page ofl-eonardot manuscripe have a gende but 

thorough integration oftext and figure, a quality rarely seor in 

c6uqi t car udtt* atkx tatn n blta t6r 6ii)+' 

lrno u ua /an'qan npu alauift qfu 

flt rt4u*tft Cdttd t4/Lttb /otrr. o, r-k l"to i;rt 

Tiin { laru ya& alla oa auhQitra., 

" 'hlknfi,ary 

[ra

Accessible Complexity: The Friendly Data Gtaphic 

Allr occasional data graphic displays such care in desiqn that it is 

particularly accesrible and open ro rhe eye. as rfthe Jesigner had 

the.viewer in mjrdar every tum 

'fhts 

wLile consrructing the graphic. 

is the friendly data gd?hi.. 

There are ma:ry specifc diferences between Giendlv and 

unfriendly graphics: 

wo.ds de spelled our, tuysrGious md 

elabohte @codins.voided 

wods m &om ieft to nshq rhe 

Bual direction to! r@dirs occidmtal 

iitde nAsases beb qdain dat 

elalohtely osded shadings, mse 

batching, dd coloa are avoided: 

iE@4 labeli .re pl,.ed on lhe sDphic 

nElf, no legod ; requiEd 

glaphic attact! view6, provotes 

coloa. ifused, de.hosd s rb,fthe 

.olor-deficidr od olq-blind (5 to 

10 !d@r of vi€w6) .d make sdF 

ofthe gnlhi. (blue @ be disrjnguisbed 

from oth€r .olox by hosr 

olor-defi.i@t psple) 

tlpe is d@, precise, modesq teHdoe 

nay be done by hdd 

t]?e is uppq+nd-lowd €r. wirh 

abbreviatioE abourd, &quinng the 

n€wer ro son tuough terl ro 

words m veEiquy, p.nj(ul,iy aloDg 

lhe Y{ns: qords m in reltnl 

graphic is crypric, requires repared 

refersces to sanded rext 

obffe .odiugs reguite soing lack 

dd forth betwem leged dd galhic 

sraphic is iepdlenr, frlbd virh 

dedgn in$sitive to coloriefi€imr 

viems; red md green uscd for 

rype n doned. ovdbedng 

t}?e i {ll dlirals, ss sif 

'With regard ro typogaphy, Josef Albers wntes: 

AESTEETICS AND TsCIINIQUE 183 

The concepr rhat "rhe rimpler tbr form ofa lener $e simpler 

1t5 reaorng w_as aD obsesron of beginning coDvmcdvism. Ir 

became somethirg like a dogma. and i, .;ll foflowed by 

moderrusrc rypographers. . . . Ophrh:lmoloey ha, disclosed 

thar rhe more rhe lerrers are difrerenLiared Goml_ech other, the 

ersrer is rfie reading. 

!irhour going irro comparisons and 

oerrrs. tr $lolrld be retlrTed $rr word\ consistine ofonlv 

caprul leners presenr rhe mo* di_6cuJr readire -b.."us.'of 

Lheir.equal heighr, equal volLrme, and. wirh ri-osr. rheir equal 

wrdrh. when comprring seriIlerters with sans_seril the ljtter 

provrde an uneasy reading.-I'he fashionable preGrence for 

sans-\eril in rexr sbows Dejther hivorical noiprrcrical 6loer lJbds, Ltela.tion oJ atu (Ns 

HN@, 1963, Evised e&rjo 197J), p.4.

184 TFxoRY oT DATA GRAPHICS 

P"oportion and Scale : Lirc Weight and L€ttedng 

Graphical elements look better together when their relative pro. 

portroh 2re in balarce. An inagraced qualiry. an appropriare 

visual linkage between the various elements, results. This musical 

score ofKarlheinz Stockhausen exhibits such a visual balance: 

In contrast, this next design n hear,ry handed, with nearly every 

element out ofbalance: the ciotted ink, the poor style ofletering, 

the puffed-up display ofa smail data set, the coarse texture of rhe 

entire graphic, and the mismatch between drawing and surrounding 

t€xt: 

?o% 

60% 

a 

a 

a 

oa 

a 

DEMOCRATIC 

l a 

j 

SHARE OF 

\Actuolrrlult: 

.I 

SEATS 

D.mocrol6 r.c.lY.d 

a I 50.9 % rot6r, 

i 55.4 % 6.ott 

5o% ---.--f\------ 

(5O% vota.. 50% !.ol.l 

aa 

4010; 

50i 60* 

DEMOCRATIC SHAREOF 

VOTE 

.L 

I

Lirres ia &ta graphics sLould be thin. One reason eighreeothand 

nineteenth-century graphics look so good is that tley were 

engraved on copper plates, with a charecteristic hair-thin lina 

The drafting pens of twentieth-century mechanical drawing 

thicLened lineworl, making it clurrsy and unattractive. 

An cf,ective aestLetic device is the orthogonal intencction of 

lines of ditrerent weiehts: 

Nearly every intersection ofthe lines in tlis desiga ftased on a 

painting by Burgoyne Ddler) iovolves lines ofdif,ering weights, 

and it makes a dilGrence, for the paintirg's cbaracter is diluted 

wittr lines ofconsant width: 

aBSTEEIICS IND TECSNIQUB 18J 

Po.tq fot th. .nibiri@ 

"MoD&ie 

ed 

N+Plalticid in Amdie," Y.lc Univdity 

An Gallcy, Ocrobcr 18 to 

Dc

i86 TEToRY oF DATA cRA?Hlcs 

Likewise, data graphics carr be enhanced by thc perpendicular 

intenections of lines of dilfernrg weights. The heavier linc should 

be a data measure. In a time-serics, for example: 

The contrast in line wcight represcntr contrait in meaning. The 

greater mcaning is giv€n to the greater Line weight; thus the data 

line should receive greater weighr than the connecting verticals. 

The logic here is a restatemert, in differcnt lan$rege, ofthc 

principle of data-ink miximization. 

Proportion and Scale: The Shape of craphics 

Graplncs should tend toward the horizontal, greater in length 

than height: 

Several lines ofreasoning favor horizontal over venical displays. 

First, analogy to the horizon. Our eye is naturally practiced in 

detecting deviations Gom the horizon, and graphic design should 

taLe advantage ofthis fact. Horizonhlly stetched time-series 

are more accessible to the eye: 

,,'.-'\/w

The analogy to rhe horizon also suggests that a shaded, high contlast 

display might occasionally be better than the floating snaLe- 

The shadbg should be crlm, wi$our moirC rffece. 

Second, ease of labeling. 

that read from left to right 

6eld: 

instead of 

lzbels 

Third, emphasis on causal influence. Many graphics plor, in essence, 

etret 

ft al 

It is easier rb write and to read worils 

on a horizontally stretched ploning- 

.- LbeL 

and a longer horizontal helps to elaborate the worLinqs ofthe 

causal variable in more detail. 

AlSTEXTICS AND TECFNIQUs 187

188 TExoRY ot DATI GlaPgtcs 

Fourd:" Tukey! couasel. 

Most diagnostic plos involve eithet a more or less de6nite 

dcpcodence that bobblcs around a lot, or a poilt spatter. Such 

olots are retha more often bener made !,i/; tLan iall. widerihan-tall 

shapes usuelly make it easicr for the eye to follow 

Eom left to rieht. 

Perbaoa tbe-most smeral euid{rce we cutr ofer is that 

s-ootdy-cfianging ;rves Ju stend being aller thau wide, 

b"t " *ilgly ciruiha& to be wid€r th; tal. . . J 

,bd, 6ndly, Play6iis example, Of thc 89 graphics in six differan 

books by-William Play6ir, most (92 percmt) ere wi

If graphia should tend toward the horizontal rather than the vertical, 

then how much so? A venerable (ffth-century n.c.) but 

dubious mle ofaesthetic proportion is tLe Golden Section, a "divine 

division" ofa line.3 A length is divided such that tle smaller 

is to the greater part as the greater is to the whole: 

a b 

-.. 

!J 

b 

b\ 

a*b 

Solving the quadratic whed e - 1 yields [, : 

In tum the colden Rectangle is 

r's +j 

2 

1616. . . 

= 1.618.... 

The nice geometry oftLe Golden Recungle is not unique; 

Birkhoffpoins out rhat at least 6ve oth* reci:ngles (induding 

the square) have one simple mathematical property or another foi 

which aesthetic claims mieht be madc:e 

lmrlNul 

r=t4r4 t=r,618 r I t.132 

Plafair favored proponions beoveen r.4 aad r-8 in aboft rwothirds 

ofhis published graphics, with most ofthe exceptions 

moving oore roward the horizonral thar rhe golden prescriprion: 

ABSTEDIICS AND TECIINIQUE 189 

3The combitutio of go@t y dd 

mFticilm s@ding the Golilen Retangle 

cm !e sm in Miloutiqe Borig 

ellieaircl\, The Coldn Nanbd &n nE 

Sciclttitt Aesthltis oJ At&ite.tur G.lew 

Yo*, r9J8) ed Tons Btu4 7t Sdek 

oJ Aftieflt C.otuttf lcop&has6. 

ecstge D. Bi*hofi ,46rr.ri, Meerc 

(C@b.idge, r9l3), pp. 2F3o-

19O TrrlORY OF D.{T.{ GRA?rrrCS 

Visual prefercnces for rectangular proportions have been studied 

by psychologists since 1860, but, even given the implausible assumption 

that such studies are relevant to graphic design, the 6ndings 

are hardly decisive. A mild preGrence for proportions near 

to the Golden Rectangle is found anong those taling parr in the 

experiments, b11t the prcGrred height/lcngdr ratios also vary a 

great deal, ranging between 

And, as is neaiy always dre case in experiments in graphical 

perception, viewer responses were found to be highly contextdepcndent.l 

o 

The conclusions: 

. If the nature of rhe data suggests the shape of the graphic, 

follow that suggestion. 

. Otherwise, move toward horizontal graphics about 5o percent 

wider than tall: 

lol have relied on Leomrd zuse, zedi 

Pd..lrior oJFoh 0{er Yo.}, 1970), 

c!. ro, for a sqtuary o{ the inmenF

Epilogwe: Designs for 

the Display of Information 

Design is choice. The theory gfthe visual display ofquantiative 

information consists ofpdnci/les thar\generare design options and 

that guide choices among options. The principles should not be 

applied rigidly or in a peevish spirit; they are not logically or rnattrematically 

certain; and it is bener to violate any principle than to 

place graceless or inelegant marks on papbr. Most princifes of 

design should be_greered with some skepticism, for word authority 

can dominate our vision, and we may come to see only through 

the lerxes ofword authority rather than with out own eyes. 

What is to be sought in designs for the display of infomntion 

is the clear portrayal of complexity. Not the complication of the 

simple; rather the task of dre designer is to give visual acces to 

the subtle and the dilficult-that is, 

the revelation of the complex.

194 rNDXX 

resthetics, gBphicrl 177 r91 

rir lolhtion 42, 1zo 

,4edd,4 ( Red fl!g") 83 

Alder. Berni 134. 14t 

Andndn Edutdtiab ia 

Aneti.ab Palitirdl Stietue Re,iew 167 

Atu"dne Statistiqre de la FnnE 167 

Anscombe's qurnc! r3 14 

area d.l q@tiry 69F73 

dnonoDicrl ghplics 26-29,154, 166 

AFes, Leoerd P. 141, rJj 

Altes, Ri.hard E. 9J 

B.mlqgd, Clam Frmds 82 

Bmooft, T, A. j3, 14o 

bzr cntn 96 97, 12612a, b9 

Brabba, Vincmt P. iJ3 

Dmeit, Vic 1r4,168 

bcfors.ftatimeseris 39 

Bcniger, James R. 20, 36 

Bdtin, Jacqles r12, 166, 169, qa 

Bi.tmore, D. P. 162 

Bidemd, Albett D. 181 

lil.tehl symetry 97 

Biachinica er Biothytid A.td ia 

Bi*ho4 George D. r39 

Slot, WiliamJ. 16 

Bonner, JolD Tyld 9a 

boring statistics 7F8o, 82 

Borisadievitch, Miloutine 189 

Boutique Data Gnphic 113 

Bowfl, William c. 9j 

box plor 97, 123, i29 

Brinton, \vilad C. 112 

BroM. DeniF Scott 116-117 

B"\ikB Week 63,83,167 

caLdtrs, gaphical 46 

Cdryonid Wata Alds rls 

Crmplcll, Angs 147 

Camplell, Donald T. 7j 

@ret 1e2o. 47, 171 

dcd m,ps 16 2oj 121 

ca!!on, r€$.rJ. 146 

.ars, frequmcy ofrepair 1?4 

Caftesim coordinats 9 

.ase andetrded 37, 47,82, 1a7 

chemical dem@ts 1o>1oj 

Chemoftrces 97,142 

Chemofl Hdmd 136,142 

chi.keD, 4,j4c!oud ?3 

.blomosorns 18173 

Clev€lmd, william S. j8,168 

color-defici@t viewds 183 

ColtoD, Farnond R. 112 

comluter graphics 26-27, 29. 42, 1\., 

116. 72e12r, 136, 153, 17a 

Connecti@t speding 74-?j 

onstant dolars 65 63 

Qoaner Repots 8a, \74 

CoDverse, Philip E. 142 

copler, coDducrivily ol 49 

Comford, Eain M. 1o9 

couty mals 1G2o, lJl 

Crotty, Wiliahj. eJ 

Daniel, Cdhbert rl3 

data-bAed labels 14e lj2 

data-built data-mss66 1j9 144 

data-ink mNimiatim 96 1oJ, 

12! 137,175, ra6, r:31 

dat. meas@6 119 1.14 

data/tex integhtion 18e182 

deodd, visul 1J3 1J4 

deflating money 6t 63 

Det Spiegel 83 

dsign mdation 6G1r3 

Dewey, ndmrd R. 1t 

Die Zet 81 

Diller, Burgoyne 13r 

distonion, gsphial 5J-J9

Doll. R. 47, 82 

dollars, consta.t 6J-68 

do!.lash-plot 1ll, rj6,1j9, 1J2 

double-fudioning labels 149F1t2 

Ducbam?, Mdcel j6 

@onomr. dat: 15, 32-:14, 33, J4-5t, 

61 64,7a.91j2, ro8, 126, 4a, 1J2, 

The Eronanbt 63,83,167 

edu.ad@L tesL', graphic 86 

Xbiobers, A. S. c. 178 

election graphie r47, r79 

electrodaphaloghm 9l 

elegance, graphicrl r77 

Uiot, T. S. ro 

erasnrg principles 96-100, 136 

*cend.e, gqphical 1l-jr 

Fi@terg, Stethen !. 14 

Finlnd, Alvr L. rt3 

Flanigm, William H. 8J 

Frd.olini, C. M. 1J3 

Ftunk_f|tt{Allseheine 8J 

Fmumeni, Joselh F. i6 

Frmcb commees 166 

Frmch winc map 2J-26 

fris.ly d.tt gh?hid r83 

fuel.cononygMphic 57-jq 

Imkhose., H. Gr.y ro, ?3,1j4 

Gab.glio, ADrotrio ?2 

gal.xies mp 26-22 154, 166 

Ccological S@ey Mps 168 

Goldenbers, Edie N. 1?o 

goYemmdt sP@ding 64 58, 1J8 

gsde scnool ghlhid 86 

Ghph ofMasicrl Parallelepipeds 68 

grid!, data-based r45-r48 

grids, white 127-129, 1j6 

Groth,ndwddJ- 26 

Gunett, Donald A. 29 

Hanlins, Tioothy H. 134 

Hamibal's @hpdgn 176 177 

Hayward, Roger 1o2, 116 

Hcrbert, George 14j 

Herdeg, Walter 80, 1J2 

hierarchy in graphiG 1jl-1t9 

high-informtion graplics 168-169 

histogr.m, lividg r4o 

horizon, an,logy to 18c18? 

Ho6e of Relrsat1tiv6 l? 

idmtificltion numbes 49 1ro 

Jatd, 

gmlhi6 in hiyeBiry entrece 

JalmeF beedc 43, 121 

JapareF ghphi6 8'84 

JASA siyle sleet ilo, 1JGiJ1, 164 

Joau.l oJ the Aneti.al Chmkdl 

Joa 

dl oJ Anen.6 Mdi.ll Astiation 

Joutnal of the Anen.q Statiai.al 

Asa.i4tion io. 1 So-t't, 164" 

Jotnal oJ Bi'logkal Chdhtry ro 

Jounat ofchehkal Phrn6 io 

JounaloJthe Rof,l Stdtini.al So.i.tf 167 

K.t!], Wilim L. 1r9 

Kellet, stai ey 9J 

Kouchoukos.Nicholas ro9 

Kuaicki, James T- 1oo, 1o9 

rNDtx 19J 

hbels on ghphics 18e182, 183, 18? 

Lambert, Johoo Hetuich 29, j2. 

L.s, Wayne 4. 129 

kw School Adftission Tesr, 

lt@ing.fim lit V.S6 rt.1r1 

I@do d: Vinci 131 182 

lits, dcf@ of ?6-?7 

Liley, P. E. 49, 1jo 

livira hi5t Dm 14o 

Long,Jobn rlniltoD 146 

Lor Angel6 smog €, 1?o 

bt Atg.Lt Tina p, @ 

lua (aco 16-20, 4? 

Me.domld-Ros,Michr€l J6 

Me.Gr€€or, A.J. 112 

tugneric monoPole 39 

mail, HoE of Rc!@obtivd 3? 

M.lcoln, An&.w H. 84 

mP, F@.h wine 2J-26 

tuP, galdi.s 26-27, 154, 166 

ma!, thcmatic 16 2? 

ma!, Yii Chi Thu 2c2r 

Mey,E.J. 3r, 34-36,40, lls 116, 12r 

Mdon, Thoms J. i6 

M,ste.ton, William 3J 

maxihitngdata-dmiry 168 

maximizing data-irl 96 1o5, r23 1!?, 

Mccl@ghan, Willian 8,

196 rNDrX 

Mcculash, MichietJ. 162 

MccutcheoD, N. Broe rd, ro9 

McKay, FErlk W. 16 

M.Pae, G.egoryj. 42. 1?o 

Mcihoe{e., H.J. 56 

Mid em dcr Rohe, Ludwig i?5 

Miller Arthu H. 

'waEd 

Miller !- 

ro 

4? I 

Miurd, ChrlesJqclh 24 2j,19, 

4e4, 57, 121, 176 r77 

Mitchell, H, L, jo 

mois: vibEti@ 1o7 112 

Molkhous, F.J. 112, r6t 

Monty, Richard A. 168, 132 

Moscow, Nalolsnt €npaign 40 41 

mo@t.in-teih+e 1J4 

multi{imctioning elements r39FlJ9 

multiple viewiilg ogles 1J4 1rJ 

multiple viewiDg delth rJ4 rjt 

mdtiwindow plob 114, 1?1 

Mulhy, ndmon84 

Ptu eedibgt o.f tha Natiogl Acadenf oJ 

Stktres, U.5.4. tro 

Psl*oloeiulBllLtin 167 

psychological expuimdts 5J-J6 

qdile llot 124, 136, 139 

FDg+liame 13o, 116, 119,149,1t2 

redudat data-in! 96-100 

relaii@lgraphica 43-Jo 

Rh6ne bridge 39 

Rickett, &meyJ. 134 

Robinion, &thw H. 20, 40 

Robyn, Dorothy L. 20 

Roge$, Alm C. 112 

Ro$, H. Lam.e 7J 

bydty, goelogy t4 It 

$mpson, RoyJ. 8J 

Serf, F. L. 29 

Schnid, Calvin I- 112 

Schnid, Stanton E. i12 

Srieu. 39. 3t, 1o9, 1ro,167, r73 

S.iefte hdir.tan, 1974 6a 

Scientfu Ahentb So, 73, 167 

Sei.feld, J. H. tF 

semi-gDphi6 178-r8o 

s@d6. Jol'n W. 168, 182 

shape, gEphial 186 r9o 

Shinohm, Miyohei 32, 8t 

Shiskin, Juliu 38 

siryocl, Holy S. r13 

siebds,3. H. 26 

siegel, Jacob S. 113 

shall multiple 42, 48, 17eqs 

So.ial Iidi.dto4, t9n 163 

sopbitiation,shphisl 8t86 

Slear, Mary llanor 81, ii2, 123 

St,tlstir4l Abdtad oJ nE Unned stat6 

$@dd-.laf plot 14o 

Stoclhausm, Kulhdnz r3a 

StoLs, Don.ld E. 142 

sMdar Tik t lImdon) 

Sphe., Wylie 139 

63 

taUe-gnphic 1aj, 1t8 1J9 

rel4ilion gn?hi6 ?6, 31 

TerldiD& IrMJ. 38 

tq.tbooks. statisti.al grlPhics 112

Thon,s, Edwrd LLwetlyn 168 

Thonas, Lavche st 

ThoDpson, Mo.rn M, 168 

Tlrower, Norhan J. w. 2:r 

rjct rnarks iz7-r29, 149 

Tilling, Laura 32,45 46 

Thz Tin6 (Lo[dan) 8!,16j 

linHsies, beforqfter 39 

thnl scheddc al, 98, 11j 116, r21 

Tdte, Edward R, 29, 53, 7t 95, rg4 

Tukey, John W. t, r14.123,12t, 

129, 136, 14o, 163, 188 

Tulet, Pd A. 114, 16s 

tfrgsren, conductiriry of lJo 

Vcnturi, Robert i17, 119, 17J 

uewing ,rchit€ctue 1J9 

Wrire., Howard 79, r42, rJ3 

Waldrop, M, Mirchell j9 

wal sndJosndl s3, 167-168 

W4:hinpofl Past 62, ?o,81 

White,X.B.8r 

white grid 127-129, 136 

Tne foliowi.s illtsiBrio ae relrintcd 6y pcrmnsio.: 

crrprER 1. An$oDbCj qn2dt c] 1973 Ameri.r Sotisbcr Asc 

oidoD Mrp orrlc C x,6. p.J/n6 F peeb,$, r.in.eron UnF 

rîn. Pado em:s.o..f Lp,,rr, Dmrtd {. cum. r, .e 

_ 

Americr Asoci*ion for tne Advecdot of scten.e. N* y ork 

Ciiy senher hi$ory 6 1931 Tho New ].ork Times Con!u,y. 

Tat on fftnled mail o 197J The New ].ort Tjhes Conr!{y. 

Mrgnetic non.pole o 1982 Anerier Asocjation for itE ^dv ce_ 

ndt of Sciece. Bndge on Rh6ft, Bibliothdque de I tco_ 

N"LiorJcde. PoDF at tru.treq to. AjCetc. !o,os, or Boo {!e1 

0 !9 4 rh. ror Ajde e. Time.. b:ed oh c:b toT \ ,eeo-yj. 

M.c-a,c-ljJoa,d t1 ttuFo.tc hnotag..J.p!n*o€;-. .ooJ 

i rus 5oos. I oado!. pf,U p! oruc, O, Srldon 

.or I .oDoTi. 

( oopeHrion "!d DAeloDn c! . p, .. Do8. o r"7o ..en.b. 

^rfr tu.In!. i H^p.LF 2. PryhcoF ro DdrraadB o .97s 

T're \.r wdl I Dei Conodb.. Dns,ng by, rv_" rqr: r ue 

.\ei rdr* \4.s,ne, h.. Ilel e.onoDr'.and.rc. , rq , n. 

p., Jit D. !-. o roTE I he Nrq 

rorr IIn6( ompan). Ol prie .979,1im-,rh..O, 

brie o 

ra ! The V "shiDg on Poj. p.J ph.e o' 

'o 

o L Auj.n*. WeeL, @ 

o MCirw-rJl, tn!. aeJprr-ofoi o raToTtFSu.Ly 

Tim6. I oodob. ccj.e 

F ofo.t o .o-q The | .ohodi., { ronl 

d govrmror c, 1977 ].ale Uriv*sit]' ?Es. New yort budget 

o j nTb" \s v.rl,l.nes.onpan).st-nlljJcdo,or,by 

liob AJen rn P{. BtuF.oir, o .4-o ttr Lo( {ngels l.n.fl. 

irldrB dour r r0/3 Tb- Wrhingbn p$r pi @ of ojt dd_ 

i,-{3 o '03. Tb- \er vorl Time, LoDos.. i Hôrsp I. Viewr 

oo c.i Dom)_" ,o3o The Ne\ york I Da Compuy. Dre of.ry 

.lc o .u-o.T\e Nes Yo.t T.hs (ohpry. -;{;r",4.Et(_ 

F@.mhJog?n 'ai 'a7r HJper a Ror. cflehdoo in c @ .uo, 

"-n,.bo 

Utu\srv PE*. pegi,orboD 

rie o rodT {.nehoD 

roln.r \ocnr A.tciri.on. Regrthrion nki o ,97o Add\ot 

WiLinson, H. R. r12, r6j 

wnson. Janes Q. 8J 

Wiskenmmn, Afthlr 17r 

Wond War I smlhic 14r 

Wumu, tuclard Saut 90 

Yn Chi Thu nap 2e2i 

Zitgrle, Nmcy H. 8J 

Zhe, leonrd 9?, 19o 

rNDXX 197 

Wtrle\ PDL.hrg I oDsnv. aegrrrior nk. o.o7o Hok, 

tubd fd wiB,on. o,u uErE. Kj-k Brld, priniru, 

uto,eNtv dd rh. ahcn, r Nrereorolog:.:l Sdi-,) di,.orr, 

t. J(lhdi, pro! ro_ & Ldmb,e,.um!r). 

"oA.i Hrvsrd dnrDg ofperod, pr.Fre ot cteDer6, -.nlr 

J. vcm.J bb. g,u.oe tufo, t"io V. 

codqd. o.o3r {h.r,d {$a:ron ro_hc Ao\seDe1, or 

\oft, -. Muluq.ldow ptor. paLt A. tuL.). B.B Du. t, p&. d _. 

I droar doo.. \ke oacJifom.r. O{!. oiptqms do Feqr..l 

]\ewsprler cont€nt @ 1979 ADelicm potitic, Sciftcc Asbcranon. 

' H{p. tr 6. Prr:llcl ! Tcr piat o .o-- Add.,on_w6.ey prb- 

!flrc \ ompa!) Bo\ ptor."rir,oh o r973 Arrh.rn \bri.-.ct 

A.o. rrioo. PLle' nsnrls o rrr AcrdeD.i. 1J6. Enprrialonu_ 

r:bw d^hb' h.D o ro-dJoll wilcy. -H\prr, 

? \-em marerf 

old o r97: T\. loq: \Lr L b \e^,w pr- t \ oC 1Dbs.a ! 

brs toiDe'. w:Cn.robydrodrDn, . o roTo Ao&mi. pE.s. Irsr 

m:p @ 0-6 PiDc.roD Ur^6i,y pts- rglo rd reoo ee, rio, s o 

r066John V/.1.) .HrF -p 3.rm.hiohmre\ 

md cdthn V'lld A.ff rmal m,JDpte o ro-i Mouron "1d 

n Dutsonr @ i!6{ Ptrcaooo prN,. I b Chror,o-oD$.Jor8cJ 

Yuni,. o ,oB) Ansidb Aso-:boo ror rhe AorrD-m;rr ol 

\o.oe Auronob i- rrcqwDcyof Jeprir o .q8z CoJM.e^ UnioD 

q oe utoFd 5ubri h!. , q^pr,R o. How o.fmr s-oupr 

voGd irr p'eidot o o3o Tte \s vo-k TimF com;dy. roE. 

4n.c o .o-o The Neq yo,( T,nc( corphy. tonr;o a, \.n, 

oBw'na. rr.rm'le 6 ro

The Visual Display

Create successful ePaper yourself

Delete template?

Save as template?