Vol 55

44 David Lewisthe pipe had a diameter of 95mm. Recordedat the west end at 12.26 mod and at the eastend at 12.22 mod. (Birch et a/forthcoming)Although there appears to be no standardisationof pipe dimensions between different conduits,presumably within a single system the dimensionsof the pipes were fixed, to aid both constructionand repair.Sections of pipe would be transported to theinstallation site and joined either by simple buttjoints or by flaring one pipe end and insertingthe next pipe (male/female joints). The jointsbetween sections would then be sealed by wipingmolten metal across the joint. The use of moltenmetal to provide the seal would have required amobile furnace to be built close to the installationsite and the construction of an elementarymould around the pipe joint, probably in thebase of the protective trench, to guide the flowof the molten metal. Clearly the construction ofthe conduit pipe would have been a slow processas the furnace and its fuel was moved from siteto site. Of the possible methods of joining pipes,simple butt joints may have been less demandingto make, but they were weak and ideally neededadditional protection. Pipes could be encasedwithin a further stone or brick housing, but sucharrangements added further expense and wereomitted where either finance was tight or it wasconsidered that the pipe was safe from damage.The Dover conduit, for example, was constructedwith butt joints protected in a stone linedconduit channel one foot square (McPherson& Amos 1931, 170). Whereas the Windsor Castleconduit was buried unprotected in land outsidethe castle, only being given a brick 'paving' onceit entered the busy upper ward of the castle,close to the distribution fountain (Tighe &Davis 1858, I, 602). A common improvementto medieval conduit systems was the subsequentinstallation of a protective housing for thepipe, to reduce the incidence of maintenanceand consequently improve the reliability ofthe supply. The Exeter conduit was relaid in astone lined channel in the mid-14th century toprotect the pipe, with the new channel beingwide enough to gain access for repairs (Holt2000, 92-3). If the pipe was not to be providedwith a housing throughout its length, thentypically critical components would be providedwith some protection. The London Greyfriarsconduit trench incorporated a marble stone, tomark both the position of underground taps andto afford some protection (Norman 1899, 259).Flared joints between sections of pipe may havebeen stronger and thus required less protection,but they suffered from the disadvantage of notproducing a smooth interior surface to the pipe.This would have made cleaning more difficult andprobably also encouraged the development ofadditional internal deposits that could eventuallycreate a blockage (Hodge 2002, 98).Conduit pipes thus incorporated two joints— a horizontal seam joint along the pipe and anend joint between pipe sections. The seam jointwould have been subject to the greatest internalpressure from the operation of the pipe, whilstthe joint between sections, although under lessstress, was potentially weak as it had to be madein situ (Hodge 2002, 314-15). Of the two joints,it would appear that the quality of the jointbetween sections was the most critical in ensuringthat the conduit system remained 'closed' andtherefore operated effectively. It is not knownhow the London conduit pipes were joined orif other external protection was provided whenthe pipe was first installed. Unfortunately, thepipes excavated at Paternoster Square did notinclude a joint between two sections of pipe,although one piece was approximately 4m longand it might have been anticipated that such asubstantial section would have had at least onejoint. A possible explanation might relate to thecommon practice of recovering (valuable) 'old'pipes to reuse the lead for other purposes. Whenthe pipe was removed it is likely to have beencrudely pulled from the soil fracturing it at theweak joint between pipe sections, resulting inonly complete pipe sections remaining buriedand unrecovered — perhaps locked in placebecause of other obstructions built on top ofthe pipe. The recently excavated pipe thereforerepresented a section that for some reason couldnot be recovered and was simply left in theground, after the pipe sections on either sidehad been extracted.Clearly, a calculation had to be made betweencasting longer pipes that would have been heavyand difficult to transport intact to the installationpoint without damaging the seam joint, andshorter pipes that would have been easier totransport but required more joints betweensections. It seems that the conduit builderspreferred longer pipes, on average 3-4m long(Magnusson 2001, 70).The standard method of construction was tolay the pipe from the source to the destination.