for finite values of N, M <strong>and</strong> R 2 . For M = R precise <strong>and</strong> rather simple formulas for thedistributions D N,N , D N,N + <strong>and</strong> for their joint distribution were {also} derived (Gnedenko &Koroliuk [60], Gnedenko & Rvacheva [64]). Koroliuk [6] determined the distributionfunctions of D M,R <strong>and</strong> D M,R + . For M = pR, assuming that p , he derived the distribution ofthe statistics D N + <strong>and</strong> D N . Rvacheva [8], also see Gnedenko [79], considered the maximaldiscrepancy between two empirical distributions not on all the axis, but on an assignedstochastic {r<strong>and</strong>om?} interval. The works of Ozols [1; 2] adjoin these investigations.In addition to the listed issues, problems concerning the mutual location of two empiricaldistribution functions were also considered. Such, for example, was the problem about thenumber of jumps experienced by the function F M (x) <strong>and</strong> occurring above F R (x). Gnedenko &Mikhalevich [68; 69] studied the case M = pR, <strong>and</strong> Paivin (Smirnov’s student) considered thegeneral case ({although} only its limiting outcome). Mikhalevich [3] derived the distributionof the number of intersections of the function F M (x) with the broken liney = F R (x) + zM + RMRfor M = R; also see Gnedenko [79].The test of a hypothesis that a distribution belongs to a given class of distributions presentsa more general problem than that of testing the agreement between empirical data <strong>and</strong> aprecisely known distribution function. The mentioned class of distributions can, for example,depend in a definite way on a finite number of parameters. Gikhman [13; 16; 17] initiatedsuch investigations by studying these problems for the Kolmogorov <strong>and</strong> the 2 tests. Atabout the same time Darling (1955) examined similar problems for the latter test.The investigation of the 2 goodness-of-fit test for continuous distributions <strong>and</strong> anunbounded increase both in the number of observations <strong>and</strong> in the intervals of the grouping isrelated to the issues under discussion. Tumnanian’s [1] <strong>and</strong> Gikhman’s [19] findings belongto this direction. We only note that this problem is akin to estimating densities.5. During the last years, a considerable number of studies were devoted to developingstatistical methods of quality control of mass manufactured goods. These investigationsfollowed three directions: empirical studies; development of methods of routine control; <strong>and</strong>the same for acceptance inspection. Drawing on these works, Tashkent mathematiciansworked out a draft State st<strong>and</strong>ard for acceptance inspection based on single sampling. Suchissues are important for practice, <strong>and</strong> several conferences held in Moscow, Leningrad, Kiev<strong>and</strong> other cities were devoted to them.Contributions on routine statistical control were mostly of an applied nature; as a rule, theydid not consider general theoretic propositions. The investigations of Gnedenko, Koroliuk,Rvacheva <strong>and</strong> others (§4) nevertheless assumed these very contributions as a point ofdeparture. In most cases the numerous methods of routine statistical quality control offeredby different researchers regrettably remained without sufficient theoretical foundation. Thus,an analysis of their comparative advantages <strong>and</strong> economic preferences is still lacking. Fromthe mathematical point of view, the method of grouping proposed by Fein, Gostev & Model[1] is perhaps developed most of all. Romanovsky [117] <strong>and</strong> Egudin [7] worked out itstheory <strong>and</strong> Bolshev [1; 2] suggested a simple nomogram for the pertinent calculations.Egudin provided vast tables adapted for practical use <strong>and</strong> Baiburov [1] with a number ofcollaborators constructed several appropriate devices {?}.A number of scientists investigated problems of acceptance inspection, but we only dwellon some findings. As stated above, Kolmogorov [129] applied the idea of unbiased estimatesfor such inspection. He assumed that one qualitative indicator was inspected after which the
article was considered either good enough or not. He then restricted his attention to the casein which the sample size was assigned <strong>and</strong> the batch accepted or not depending on whetherall the articles in the sample were good enough or at least one of them was not. As a result, anumber of accepted batches will then include defective articles. The main problem here wasto estimate the number of defective articles in the accepted, <strong>and</strong> in all the inspected batches.Kolmogorov provided an unbiased estimate of the accepted defective articles <strong>and</strong> concludedhis contribution by outlining how to apply his findings. In particular, he indicated theconsiderations for determining the sample size.Sirazhdinov [13; 17] methodologically followed Kolmogorov, but he considered a morecomplicated case in which a batch was rejected if the sample contained more than c rejectedarticles. Interesting here, in this version of the problem, is not only the estimate of anadvisable sample size, but also of an optimal, in some sense, choice of the number c. Theauthor offered reasoned recommendations for tackling both these questions.From among other contributions on acceptance inspection, we indicate the papers ofRomanovsky [112; 122], Eidelnant [12] <strong>and</strong> Bektaev & Eidelnant [1]. These authors werealso influenced by Kolmogorov.Mikhalevich [4; 5] studied sequential sampling plans. He also restricted his investigationby considering qualitative inspection when the acceptance/rejection of a batch depended onthe number of defective articles in the sample but the quantitative information on the extentof overstepping the limits of the technical tolerance or on other data important formanufacturing were not taken into account. Mikhalevich’s method of studying was based onWald’s idea of decision functions (1949).In our context, this idea is as follows. To be practical, the method of inspection should beoptimal in a number of directions which are to some extent contradictory. First of all, theinspection should ensure, with a sufficiently high reliability, the quality of the acceptedbatches. The cost of the inspection should be as low as possible. Then, the choice of the mosteconomical method of inspection should certainly take into account the peculiar features ofthe manufacturing <strong>and</strong> the nature of the inspected articles. It is therefore reasonable toassume as the initial data the cost of inspecting one article (c); the loss incurred whenaccepting a defective article (a); <strong>and</strong> the same, when a batch is rejected (B).Suppose that a batch has N articles, X of them defective. Then the mean loss incurred by itsacceptance isU X =ma(X – m)P[d 1 ; m|X] + BP[d 2 |X] + ckkP[ = k|X].Here, m is the number of recorded defective articles, P[d 1 ; m|X], the probability that the batchis accepted <strong>and</strong> m defective articles were revealed from among those inspected; P[d 2 |X], theprobability that the batch is rejected; <strong>and</strong> P[ = k|X] is the probability that the decision ismade after inspecting k articles. If the probability of X defective articles in a batch is (X),thenNu =X = 1U X (X)should be considered as the mean (unconditional) loss. The optimal method of inspection issuch for which this is minimal.Mikhalevich studied optimal methods of inspection assuming that the size of the batch waslarge <strong>and</strong> that consequently the hypergeometric distribution might be replaced by thebinomial law. Optimal here were certain repeated curtailed samples. A number of
- Page 4 and 5:
[3] I bear in mind the well-known p
- Page 6 and 7:
successes of physical statistics. B
- Page 8 and 9:
classes of independent facts whose
- Page 10 and 11:
distribution is a corollary of the
- Page 12 and 13:
examine in the first place the curv
- Page 14 and 15:
12. According to Bortkiewicz’ ter
- Page 16 and 17:
generality, the similarities taking
- Page 18 and 19:
one on another, as well as the corr
- Page 20 and 21:
is inapplicable because the right s
- Page 22 and 23:
Instead, Slutsky introduced new not
- Page 24 and 25:
abandoned in August 1936, but it is
- Page 26 and 27:
last decades, mathematicians more o
- Page 28 and 29:
charged with making the leading ple
- Page 30 and 31:
motion and a number of others) are
- Page 32 and 33:
phenomena. It is self-evident that
- Page 34 and 35:
Such new demands were formulated in
- Page 36 and 37:
The addition of independent random
- Page 38 and 39:
automatic lathes, etc. Here, the ma
- Page 40 and 41:
11. Kolmogorov, A.N. Grundbegriffe
- Page 42 and 43:
period 1 and remained, until the ap
- Page 44 and 45:
of the analytical tool rather than
- Page 46 and 47:
with probability approaching unity,
- Page 48 and 49:
logic. The ensuing vagueness in his
- Page 50 and 51:
2. Gnedenko, B.V. (1949), On Lobach
- Page 52 and 53:
will be sufficient, although not ne
- Page 54 and 55:
nlimk = 1P(| k (n) - m k (n) | > H
- Page 56 and 57:
favorite classical issue as the gam
- Page 58 and 59:
and some quite definite (not depend
- Page 60 and 61: influenced by a construction that a
- Page 62 and 63: P ij (1) = p ij (1) , P ij (t) =kP
- Page 64 and 65: 4.2d. Bebutov [1; 2] as well as Kry
- Page 66 and 67: are yet no limit theorems correspon
- Page 68 and 69: described, from the viewpoint that
- Page 70 and 71: conditional variance and determined
- Page 72 and 73: Romanovsky [45] and Kolmogorov [46]
- Page 74 and 75: Let S be the general population wit
- Page 76 and 77: Part 1. Russian/Soviet AuthorsAmbar
- Page 78 and 79: 2. On necessary and sufficient cond
- Page 80 and 81: Gnedenko, B.V., Groshev, A.V. 1. On
- Page 82 and 83: 52. ( (Mathematical Principl
- Page 84 and 85: Kozuliaev, P.A. 1. Sur la répartit
- Page 86 and 87: Obukhov, A.M. 1. Normal correlation
- Page 88 and 89: 30. Généralisations d’un théor
- Page 90 and 91: 22. Alcune applicazioni dei coeffic
- Page 92 and 93: 10. A.N. Kolmogorov. The Theory of
- Page 94 and 95: Kuznetsov, Stratonovich & Tikhonov
- Page 96 and 97: In the homogeneous case H s t = H t
- Page 98 and 99: to such a generalization. He only s
- Page 100 and 101: In the particular case of a charact
- Page 102 and 103: as it is usual for the modern theor
- Page 104 and 105: 1. {The second reference to Pugache
- Page 106 and 107: Smirnov, Romanovsky and others made
- Page 108 and 109: determined the precise asymptotic c
- Page 112 and 113: Mikhalevich’s findings by far exc
- Page 114 and 115: Uch. Zap. = Uchenye ZapiskiUkr = Uk
- Page 116 and 117: Khinchin, A.Ya. 43. Math. Ann. 101,
- Page 118 and 119: 7. DAN 115, 1957, 49 - 52.Pinsker,
- Page 120 and 121: Anderson, T.W., Darling, D.A. (1952
- Page 122 and 123: Statistical problems in radio engin
- Page 124 and 125: observations for its power with reg
- Page 126 and 127: securing against mistakes (A.N. Kry
- Page 128 and 129: of the others, then its distributio
- Page 130 and 131: In Kiev, in the 1930s, N.M. Krylov