THE ENIGMATICWATER-MILL

by tsradf<,rrdB. Blaine

Marc Bloch,writing ncarly forty yearsago, brilliantly cornprcsscd into a brief articleenouglt documentation to suggestthat the water-powcrcdgrain-mill bccarnc a basicleature of Medievallile; his findings pointcd,ntorcovcr, to tlte applicationot'hydraulic cncrgy to variousother industrialprocesses Sluring the MidtllcAges.l Subsequentto Bloch'spioneering efforts, Bertrand Gille2 and R. J. Forbes3hou. thrown wide nets,adding extcnsivcly to our invcntoryof mcdicvalwater-powcred devices, while detailcdregional studies such as those of A. M. Bautier,4Germain Sicard,5 and E. M. Carus-Wilson6 havcrevcalcd clear indications of local industrialdcvclopmcnt related to the ernploymcntof hydraulic mcchanisrns.Lynn Whitc,moreover, includes thc explorationand uscsof water-puwerarnong thc technologicalfactors wlrich hc recognizesas critrcalin the forcesof attitudinal,/social,u and economic9changc during the MiddleAges.

For all its known incidences,howevcr, and for all that we suspectconcerning its importance,therc is an immenseanonymity rcgardingthe medievalwater-mill. Unccrtainticssurround its genesis, diffusion,and anatomies,and thcsein turn encumberthe searchfor new data and for furthcr insightsinto its significancc.To illustratcthe nature and exlent of the problem and to give somc usefuldirection to thc currcnt enthrrsiasmfor the subjcct,thc presentpaper probesthree specific dilemmasconcerning the historyof the water-mill.l0

l. A Problem of Origins

The first known attempt in the West to harnessthe energyof running water for practicalindustrial purposesreflects in the appearanceof the water-poweredgrain-mill in the late first century B.C. This is not to deny the contemporaryexistence of relateddevices. The norfu, for example,a vertical wheel fitted with paddlesand buckets,drew water for irrigationfrom the current which turned it.1l And Greco-Romanengineers conceived a variety of hydraulicautomata in which runningwater turned wheels;the sole objective,however, was the turning itself or the noise thus produced,l2 of little practicalvalue except as temple tricks. In none of thesewas there the transmissionand use of the power available. With the grain-mill, however, the energy was indeed transmitted from the wheel and utilized by a grinder. The earliestcertain Western evidence of this comesfrom the writingsof Antipater of Thessalonica,Strabo, and ,as well as from somecarefully studied archaeolosicalremains. 164 B. B. Blaine

Alr epigrantattributed to Antipalcrannounccs the cmanciputi()nof'nraitlservants frt-rnr the drudgcry of hand-rnilling: Ccascliont grinding.yc womcll who tclil:rt thc rnill; slecplate, cven if'thc crowingcocks anncluncc the dawrr. For Dernetcrlr;.rs ordercd thc Nyrnplrsto to performthe work of your hands,and thcy, lcupingdown on top of tirc whcel,turn its axlc which,with its revolvingspokcs, turns tlte lreavyconcavc Nisyriantnill-stoncs. We tasteagain thc jclysof-tlrc prirnitivclil'c. learning lo fcaston the pr

Thc sanrcsort ol uncertainticssurrountl llte h.t,draletuwhrch Strabr) says was forrrrdat Cabirl, Pt>ntus,on thc LycosRivcr nearlhe palaceof Mithritlatcswhen the placetcll to Pompo,in (r5 tl.C.20 If thc cleviccwus built with the palaccin 120 B.('., it woulclantedttc Antipater's rnill by ,n.uy y..rr.2l Thcre'is,howevcr, no evidcnceol'this; thc earlicstdatc'that cart rcasonably be assignedis 65 B.C., whcnthe rcportedevent ()ccurrcd, althouglr the salestestimatc is ca.7 B.C.,at whichtimc Strabo was travclingand writing the Gu1raphiu.'- lt is impossihlc,nrorcovcr, to fashionfrt>m tlrc rcporl any kind of positivedcscriptiorr.zr

Vitruvius,a Rornancnginccr writing no laterthan l4 8.C..is lhc first to providctcchnical dctail. llc clrel'ullydescribcs a verticalunclcr-shot nrill, includingthc-csscntial gearing.24 Gearirrg harl bcen r) known in the HcllenistrcWorld sincc the third ccntury'8.['., brrtVitruvius's mill repre.scrrtsits first inclustrialuse antl rrtarksa clccisivcstep in the devclopnrcntof'powc-r nrcchanisn,s.26

('oncurrcntwith thescevents in tlte MeditcrraneanWorld wls thc probableappcarlncc in Scarrdinavia of the horizontalgrain-mill. Excavations at Bolle,Jutiand. revcal traccs of two dam sitesdalirrg frorn tltc beginningof thc Christianera. lncludedin tlqescis clcarcvidence ol the danrs.rcscrv()rrs and races. Carefulanalysis of silt depositsindicates, moreover, not only rnillsbut alsothat thc1, werc of the horizontaltype."'

The earlysimultaneous appearancc of the verticalwater-rnill in Romansourccs, lirmly establislred, and of the horizontalmill in northernEurope, strongly indicated, corncidcs with the earliest reportof water-powerat work in China. JosephNeedhanr shows conclusivcly that hydrauliccncrgy wasin use thereby 20 A.D. to driverice-hulling trip-hamrners and by 3l A.D. to work rnetallurgical bellows. Althoughillustrations are lackinguntil the early lourteenthcentury, his analysisof thc total evidenceindicates that the horizontalwater-wheel was used as a motive forcc from the t'irst The EnigrnaticWater-Mill 165

ccnturvonward, as with thc bellows,and that the verticalwatcr-wheel, though lcss common in China, datesfrom a comparablyearly pcriod, espL'cially in associationwitlr trip-hammermechanisms, whiclt are lnore intclligiblein that cotrtext.rd Ncedham,moreovcr. provides evidcncc from Chinesesources ol-yct anothcrhydraulic device, a watcr-poweredspoon tilt-hammer (iill-and-spill mechanism), again not ill,rstratecluntil thc fourteenthccntury but mcntionedin tlre literatureby the mid-twelfthccntury and probablyof greatcrantiquity.29 Onc wonciers,in lact, how far back into Chinesehistory tltis gocsand tltc cxtent to which it may havccharacterized thosc first trip-hammers,or cvcn the bellows.

Clearly,thc point ol originof the conccptof watcr-drivenmachincry rcmains problematical. Vastly^ separatcareas and threediffercnt kintls of mechanistnare involved. Argunrentsciting either Chinaru or thc Grcco-RomanWorld3 I abound,but they remaininconclusive. Engaging but equallyundemonstrable at prcsentis thc suggestionrrthat somemiddle ground may be criticalhere; that is, sornemobile Eurasianpcoplc hit upon the conccptof water-powcredmachincry and was the instrumentof its dil'fusion.various adaptations being effccted wltereverthc idea was reccivccl..A final possibility, arguablclrom thc prescntstatc of the cvidcncc,is that therewcre severalvirtually simultaneous invenlions: the vcrticalmill-an effcctivcconrbining of rccentlydcvcloped rotary grindstoncs,gears and the noria in thc easternMcditerrancan, the horizontalmill in northernEurope and in China, and the fill-and-spilldcvice in E,astAsia.

2. A Problem of Predominance

Evidenccof tllc water-powcredgrain-mill in tlic Westaccumulatcs from the time of Vitruviusonward. Sincc,howevcr, so little ol'the testimonyis descriptive,consisting rather of casualreferenccs or cntries in property inventoriesand r.ltherlegal documents,Jrit is not clear which of the basictypes of grain-millwas preferrcdand/or most extensivelyused. The total relativelyunambiguous evidencc of the verticalundershot, vertical overshot, and horizontalmill for over a milleniumcan be organized into a lew brief paragraphs.

With respectto the undershotarrangement, our carliestdescription is that of Vitruvius,discussed abovc. Next, relatingto the tragiccataclysm of 79 A.D., Italianarchaeologists have reconstructcd a water-wheelfrom volcanicash impressionsfound at Pompeii. Piecesof the cast and the reconstruction are in thc MuseoNazionale in Naples. The wheel,having eighteen paddles set betweenparallel rims, is of thc vertical typ..34 Our first picturc of any water-millis includedas part of an early fifth- ccntury mosaicin the Great Palaccof Byzantium. None of the internal mechanismis shown: merely the wheeland thc mill-house;but the wheelis standingvertically over the water.r) No vertical undershotmill appearsagain for nearly 800 years;a late twelfth-centurymanuscript of L'lmage du Morule,by Gautier of Metz, containsthe externalview of a mill-housewith an undershotwheel,36 as doesa thirteenth-centuryminiature in the Vieilrentier.ST Our first indisputableillustration of the gearing^associatedwith an undershotmill appearsin a South Germanmanuscript dating from ca. l205.iu A late thirteenth-orearly fourteenth-centurySpanish reliquary provides internal detail of an undershotmill, includingthe water-wheel,axles, gears, stones unO ttopp.t.39 Verbal descriptions40 and pictures4I of the deviceare numerousfrom the fourteenthcentury onward. t66 B. B. Blaine

Useof' the overshotvertical mill can be firmly docunrcntcdin the Romano-Byzantineworld by thc rnid-fifthccnlury, and thercarc indicationsof its developmcntlhere evcn carlicr. The uninrpeachablc evitlcnce,supplied by arclraeology,is that of an overshotmill in thc AthcnianAgora dating frorn thc regionof Leo | (451-414).*' Originatingconsiderably carlicr, pcrhaps, and only slightlylcss securc irr its precisehistorical and technicalveril'ication than thc Athenianexample, is a largemilling installatiun at Barbegal(near Arles)which seemsclcarly tcl haveernploycd several overshot whecls. I have visitedthe remarkablywell preserved site and find no quarrclrvith FcrdinandBenoit's classic analysis of thc cxcavation.43Waters fronr the aqucductLes Baux weroclivertcd into two nrill-raceswhich werecaused to tumbledown an inclincconrposcd of a descendingseries of eightlcvcls. At all cight lcvelscach strcam supplied the power for a set of'mill-stones,totaling sixtcen nrills. Tlrc arrangementis most intclligiblein termsof over-shotwhccls, and coins,nredals and pottery indicatc that this installationdates from as earlyas the late third or beginningof thc fourth ccntury. Therc- al'tcr,as with tlrc unde-rshotdevice, dcscriptive evidencc of lhc t>vcrsholnrill doesnot appcarfor U00 ycars. A thirteenth-centurynranuscript, the Liber Artsclni rlui dit'imr Apologeticturr,contains a skctclr of an overshotwheel supplicd from abovcby a trough.44 Furtherreprcsentations accunrulatc front the fourteenthccntury ,rnward.45

Thc simplicityof thc l'rorizontalmill encouragesthc assul.rptionthat it was thc carlicst46or tlle most primitive4Ttypc, and as notedabovc, thc hydraulicgrinders of Antipatcr.Strabo and Pliny havebccn variouslydescribcd as ernployinghorizclntal whecls.48 As a nraltcrof fuct, howcvcr,firnr descriptivc evidenceol'thc horizontalmill cluringthc latc classicaland nredicvalperiods is exccptionallvrlrc. Within the Westcrncontext, the first inclicationol-the horizontaldevicc, based upon somcthingnlorc tharrhistorical conjeclure, comes from the Jutlandexcavation and doesnot appearto antedatcthc late first-cc'nturyB.C. vc'rtical mill of Vitruvius. Our earlicstlirll dcscriptionof the horizorttalmill is containcdin the arrcicntIrish law code..Sorr'ftrtsMrr; herc tlrc severalparts of the cornpletcrttill are ljstcd.49 F.lenrcntsol this cotlcnray date lrom the rcigrrof King Laigheirel43S-44-l), but philological considcrationsplace the'major compilation in thc latc ninth or c'arlytentlr ccntury.)u Ncarly cnntcrrrporancouswith ScnchusMor is a verbaldcscription by (-ormac,bish<,rp of Cashel(d. 903), which se_e.rnsclearly to contemplatea horizontalmill: hc likensa whirlpoolto thc whirlingof'mill- paddles.)r And the clcvcnth-ccnturyepic TttgailBnritlne Du Dergalists the parts of what appcars to be a horizontalnrill.)r Tlrc next evitlenccrrrescntlv available of thc horizontalwheel does not occuruntil the fiftcenthccntury.5S l{.r. it appelrslir.t. r, thc powersource for what looks to be a wooden-pipeborer, in a Germanmanuscript of ca. I437" arrdsliglttly latcr, in its more familiar grindingfunction,_in thc notebooksof two engineers,Mariano di JacopoTaccola (MS datablebetween '144 I and l45S)55and thc so-calledHussitc Engineer (MS datablebetween ca. l4l2 and ca. 1485).s6 From this point onward the horizontalwheel emergesfrom the documentswith ever greaterfrequency and sophisticatcdusage.57

Clearly,the gearcdvertical device was the most eificient. For one thing, sincegearing obviatecl thc necessityof placingthe stonesimmediately above the wheel, the wheel could be positionedso as to take best advantageof the current;also, the mill could be locatedanywhere with respectto the water, thus faciliatatingaccess to the mill-houseand permittinga much largeroperation. But of particularsignificance is the fact that gearingpermitted power control, making possiblethe adjustment of the speedof rotation. The verticalundershot wheel did not alwaysnecessitate a millraceor chute, though thesewere often included,particularly where a millpond was used or the main current inadequate. The verticalovershot wheel, however, required some structuralprovision for feedingthe water from the The EniematicWater-Mill 167

top, but this encurnbrancewas compensatedfbr by the addedpower suppliedby thc weight of tlre watcr itselfon thc wheel. Whichof thesetwo vcrticaldcsigns predominated cannot prescntly be dctcrnrined.A. P. Usher'scount of thc total illustrationsknown to hirn up to the sixtcelth century lcadshinr to suspcctthat thc undcrslrotwhccl was the morc commonof the two forms.)u Although the availableevidcnce' does not warrxnta rnajorcriticisnr ol Ushcr'sconslusions, it sh<-ruldbe notcd that nranyof thcscpicturcs arc simplccxterior views ol'mills as part ol thc landscapc;an artist mighl be inclinedto renderan underslrotwheel hastily and ornit any complicatedovcrhead structure.

Thc greaterefficiency of thc verticalnrill neednot, however,provc its predominantuse, l-or associatcd with that vcry quality wasa greatcrcornplexity and inevitablyhighcr cost of constructionand rnaintcttance.We shouldnot, of coursc,bc surprisedto learn,il'we could,that in tlrc clcvcnthccntury thc mills ol' thc important Abbcy of Viecht (Tyrol) employedvertical whcels to receivethe watcr frotn a conduit which llcnry IV had cut lbr thc monks,at grcatcxpense, across the faceol a rock cliff.s9 Likcwisc,howcvi-'r, wc shoulclnot be surprisedto lcarn,if wc could,that somcof the 5,624 or more watcr-rttillsin severalccluntics of'Domesday Englantl60 werc of thc horizontaltypc this is in spiteof the fact that no medicvaltraces of the devicehave as yct becnfound there,but due to thc fact, as shownabove, that horizclntalwhcels had long becnat work just acrossthe Irish Sea. The carlierwide use of thc horiz-ontalrnill might be further indicatedby the cxtensivcevidence of survivals in earlymodern timcs, and eventoday in somccases, throughout northern Europc, e.g. in Scandinavia, Ircland,the Isle of Man, the l{ebrides,the Orkneys,the Shetlands,thc Faeroes;throughout central, scluthernand eastern[:urope, e.g. in the Alps,Francc, , Italy, Bulgaria,Yugoslavia and Greece, and in the Holy Land.6I

The temptation to build modelsis irresistible.One might picture, for example,a medicvalEurope with the efficientvcrtical mill, prcsumablya Romaninvention, prcparing flour for the technically more sophisticatedartd tnore afflucnt populations in the busy manorialand urbancenters of lifc. Among lessskilled and lcssafflucnt peasantcultures, meanwhile, particularly in high and ruggcd placeswhere fast-running mountain streams provided the appropriatekind of energy,the horizontal mill usuallypredominated. Originatingin northern Europeand diffusedperhaps by wide-ranging Norsein their travelsfrom Greenlandto Byzantium,or originatingin t'hina and flowing through Europe from a southeasterlydirection, or both, the horizontalmill performedits humble serivce for generations,then to bc discovcrcdfor its latcnt turbinc qualiticsin thc fifteenth and sixtccnth centuries.62Given the informationallacunae, however, such a motlel is usefulonlv if usedto launch furtltcr inquiry, not to closeit.

3. A Problem of Design

The practicaluses of water-powerbeyond the grindingof grain c:uneto be recognizedduring the Western MidclleAges in severalvital industries.63 For purposesof the presentdiscussion, thrce specificexamples can be taken: the preparationof beer mash in the food and drink industry, the fulling of cloth in the textile industry, and the activationof forge-hammersin the metallurgicalindustry. From a technical standpoint,what theseparticular examples share in common is that eachof them seemsto have employed 168 B. B. Blaine

pourrdingmecharrisms, which reprcscntsl crttcirtladaptl(iort ol- tltc watcr-rvltt'cl. (-hamctc-ristically, hlwcvcr. wc havoyct rnuch to lcurn abotrt this ltdaptlrticn.Wr'arc not cvcll ctrtlirr of rts csserrtial dcsign,as a briet' survcy ol thc known l'actssltow.

An architccturalplln preplrcd cl. 820 lirr (lozbcrtus.lhbot ol'St. Clll (lJl6-ll-ltr).64includcs scveral rvorkslrt'r1.rs,c.g.. ol' thc llthc opcrators (tttrtnlttrt's), tlrc tlrrncrs (ctriarii), the goldsmitlts (utri.lit't'.s), tlre irorr-srniths(Jahri f crrarrrt'rttttrtun)lrntl tlre'tirllers (l'trlhtrtcs). Notlring lrcrc indicatestlrc rrsc of rvater- i)owcr or of ntcchaniz-cdporurtlcrs. Next to tlrc'workshol) arcl. Irowevcr.arrd locatcd on tlrc outcr cd.rtr'oltltc nronasticcorttplc'x, are two cnclosurcs.orte contlrininge skctch r>f lwo whcelsancl labclled trutlac. Ihc o{lrc'rcontainirrg r skctc}rof'two lrammer-likcobjcr:ts. or wliccls witlr lltnrncr-likc cxtcnsiorrs, arrtl labcllcd 7rilac. I)irc.ctlvintcrior to thcsc cnclosurt'sis l lurgcr orrc. itlntlinirrg I brcwcry //ui' lrutribus ('()tl'iutur tcnJisa) dircctly adjaccnt to thc Pllnr', urrd a bakery (pistriruutt) directly adiacent to lltc nutluc. Altliough prclol is lacking, tw() cnrillcntly relsonablr'conjcctrlrcsurc possihle: lirst, tlrc pildc teprescntwatcr-l')owcrcd pourrders; second, thcsc lryclruulicpounclers wcrc uscd to preplrc be'cr 'Thc nrash.a lr)orc rppropriatc wav o1 pulverizingrnllt llirn bv grirrding. nronks ol'St. Ciall. ti>r whorn thc plrn prcsunrablvwas irrtcndcd,or livg_clurniclsl a long trudition of water-pt)wcrr'dgrain-rnillinr, goirrg back at le'astto the nrid-cislrthccnlury.ob Ihis tradrtion.cntirclv consistcntwith the gencralpatlcrrts t>l-carly rncdievll techrtologv,lcavcs littlc scriousdoubl tlrlt Ihc rttoloc rcprcscr)lwrlcr-drivcn !lririn- nrills; it strongly sllB,g,csls.nr()rcover, tllat thc pilac, Ioctted lrcxt to Ihe ntoluc alons arr r-xterrorsidc ol'lltc cortslruclion,wc-rc vicwcd h1'thc urclritt'ctas hydrutrlicporrrrtlrlg rncchunisrus 1o lrc powercd by thc sanlc sourcc. A tkrcurrrcrrtclatablc ltg7-q0-+ arrd associatctlwitlr St. (lall ttrcrttions"aquis 7 l(lr.rrnlnlquctlccrrrsibtrs. aut cllusrrris.rrrolinis vcl pilis."h Tl,ir tert strorrgll,inclicate s watcr-powe rcd ltrturtderswithin thc purvicw ol'the rnonksof'St. (iall hv cu. 900 arrdundcrscores tlrc plausibilitl' that sinrilardcvices werc contelnplatcd by thcrr lrchitect ciglrty ),carscarlicr. Morcovcr, the relrtionship in the Kktstt'rpbtr of thc pilac to tlrc brewerv suggestsllrat thc poundcrs wcrc intendcd for thc nraking of becr rnash.6E Nor can u tradition of watcr-powcr in thc brcwcr's art by this datc bc cl<-rubted.for arr act ol- Charlcs thc Bald of 8(rI rcfers to "rrrolcrrdilturnulrunl cunr carnbJ supcrpositl."69 And tlr0 cvitlcnccaccunlulatcs frorn thr- nirrtlr ccntury onwlrd untlc'r thc usual dcsignalionrtt

Wrereas thc-rrormal utilizltion ol'pounders, as distinguislredfrorn grindcrs,crn tlnly be assumcdwilh respectto thc prcparationol'beer mash, they must ine'vitablyhave been employcd in the lnechaniz.ed lllling proccss. Fullirrg-rnillscalnc to be used for the cleansingand thickcning of newly wovcn cloth, thc tcxtile being soakedtnd poundcd in an appropriatesolution.. We hrve two rncticulousrcgional stucliesof thc nretli0valfulling-mill: that by Bauticr lor ltrance'71and by Carus-Wilsonfor trngland,T2"" botli of which docunrr'ntits subslantialimportlnce by the thirtcenth century. The docurncntiitionis fiterary and conrplicatcdby the occurrenccof sevcraldiflcrcnt tcnls, e.g. Jitllatttrilutr.gauchatoriurrt (ltrlian gualt'heria, Gernran llalk<'rci), parat()rium (French usittc o !)0ru), atld balut()riunt. The lrst of these terms actually specifiesthc pounding rrction neccssarilyinvolvcd.'' but nonc of the words. nor their contexts. bctray the designof the pounder. This is as true of the carliestcredible indication of the dclice. the "ntolenario et batedorios" frorn the abbey of Saint-Bernardde Romans (Dauphind) ca. 990,74asit is of later appcarances,e.g. the "molendinunr vcl paratorem"T5 and "molendinunr fullonarium"T6 in the eleventhcentury and the "molendinunr draperiunr," "rota draperia," antl "molendinurn ad operandum pannos" in thc twelfth century." The EnigmaticWater-Mill 169

Thc utiliz-atiolrof water-powerin thc medievalrnetallurgical industry clearly was known hy thc carly thirtcenthcenlury. ln 1224,ior cxantglc.the villageof Toakcr in }lalland(Swedcn) had an iron-rnill: "de rnolcndino,ubi fabricaturfcrrurn."/6 Thc wordsdo not indicatcthc meclrarrisrninvolvetl. Until the advcntof the blast-furnace,with its productionol' intenselreat through thc usc of' largclrydraulic bellows,which did not takc placcbcfore tllr. late fourtcenlhcentury, wc call cnvisionin suclrirorr- mills as that al Toakcrrclatively sntall water-powered forge-fire and/or smclting bcllows antl pounding tnecltattisms.T9Thit is to assuqrc,oI course.that generalrefcrences to iron-rriillstlo not conternplatc l sharpeningand polishingmills80 or hydraulicmachinery associated with rnining,8 both of wliich can be disccrnedduring the period. Thc presentdiscussion dcles not involvetlre bellows,except to nore thal the ticvicchacl, in fact, becnmcclranized by the thirtcenthccntury.82 Ol'direcl rclcvancehere is thc trccttrrenceof the termsnwlleus anrJ,lurttmcr in tlrirteentlr-cerrturydocunrcnts relating to tlrc importantiron industryin the Obcrplalz.8SNor can therc by any seriousdoubt that hydraulic forgcshad appcaredeven earlicr. For examplc,in the Obcrpfalzitsclf, thc town ol'schrnidntiirrlcn, locatcdon thc rivcr Vils. trrccsits narncback to no later than 1028,84suggcsting that thc arca was alrcadythcn distinguishablefor its watcr-povreredforges. And in I135 wc cncounter"rnolcndinurn unum et stanfunum" belongingto the nronastcryof Admont in the iron-producingregion ol Styria.85 ln nonc-of'this, howcver, js therean indicationof thc dcsign.

Tlrc cntploymcntof liydraulicpoundcrs in thc rncdicvalfulling and mctallurgicalindustries, antl cluitc probablyin the art of brewing,cln be shown.but the technicalcletails catrnot presently bc dcmonstratecl With considerablecertairtty, howevcr, thc fie'ldol possibilitiescan be narrowcd. Thcre is no evidcncc. fcrrt-xamplc, of rcciprocatingspill-rnd-fill rnechanisrns in rnedievalindustry, such as thoseref-crred to abovcwithin the Chincsccontcxt. Nor did compoundcranks and conncctingrods, by which lranslationcan bc madefront rotary to reciprocatingnrolion. appcar in-L-urope trntil the fifttcrrth centtrry.S6tsy elimination,ancl consistcnt with thc known patternsof medievaltechnology, rrrc poundingactitln involved here was produced by camsset into tlte horizontalaxlc of thc verticalwater- whecl. The cant itself,like thevertical wheel, datcs back to classicalantiquity.8T Thcir rnlrriage, lrowcver,scerns not to haveoccurrcd prior to the ritcsperlitrnred by thirstylnonks in thc c'arly ninth centuryor by textileand metalcraf'tsmen around thc turn ol thc eleventhccnturv.

But the mattcr still rcmainsunresolved. Thc fact is, canraction can be employcdirr thc clperation of two very differenttypes of pounder. One of thcscis thc vcrtical-b€arnstamp; the othcr is thc recumbenttrip-hammer. As we have noted, the tcrms hsmer and stanf do appearin certain regional records,but thereis no necessaryrclationship betwecn these words and thoscused by me hcre to distinguishbetwcen the two designs.Onc searchesin vain for illustrationsprior to their appearance in tlte fifteenth century. Ironically,when at last they first occur, in the notebook of thc sanrc anonymous"Hussite Engineer"who hasp^rovided us with one of the earliestrenderings of a horizontal rnill, both typesof pounderare ol-lbred.uuWhich of thesepredominated in thc rcvolutionarvindustrial developmentsof the previousfew centuriesin Buropecannot presently be affirmed.89

Severalpotentially fruitful areasof inquiry regardingthe mattersdiscussed here are evidenr. The problem of origins,for example,makescritical all new revelationsof hvdraulicdevices in Eurasia r70 B. B. Blaine

from bcfurc thc lirst centuryB.C. through the first centuryA.l). Of particularirrtcrcst is thc kind ol mecharrisrninvolved, its provcnance.and the agentsarrd pattcrns of its difhlsion.

Thc rluestionof predominanccdrarnrtizes thc generalrcquirctnertt of morc descriptiveinformation rcgardinggrain-nrills throughorrt the MiddlcAges. Specificallynccded is directevidcncc of vertical undcrshotand verticalovershot nrills frorn the fifth 1o the late twelfth century,prcsently a massivc- gap ol 800 years. t:qu.rllyncccssary is cvidcnceof the ltorizontalnrill in E.uropcpriur to lhc fiitcenth century;at the rnontcntonly Jutlandancl lreland testify to its earlieruse.

Thc particularproblem ot'dcsigrr cited herc has two importantdimcnsions. The l'irstof thcseis thc unccrtaintyregarding the placcarrd tinre- of tltc crucinljoinirrg of tlre water-wlrccland carn;the most likely possibilityto datc is thc Alpinc rcgionwithin a tinrc-franrcfrorn thc carly nintlr to lhc late tcnth century. .Secorrdly,wc tlcl not lrow know thc spccificdcsigrr ol'tlrc poundcrsinv<;lvetl; the scvcralrelevant industrics, i'.g. brewing. tcxtiles and nrclallurgy,nccd to be searchcdfbr clear pre'-fiftee'nth-ccnturytraces ol vcrtical-bclnrstarnps and/or rccurnhcnt trip-harnrncrs.

Thc prcscntcnigmas atte'rtdirrg tlrr'lristury of thc watcr-rrtilldo not corrtradictwhat scvcraldccades of scholarshiphavc already shown concerning thc prevalenceof the dcviceduring thc MidtllcAges and its placeilr thc tecllnologicalirrgrcdient ol'our culturalmatrix. lrurlhcradvanccrttcnt of thc subjcct,howcvcr, rcquires a concentrationon kirrdsof problcnrsoutlirrcd here.

NOTES

l"Avcnenrcnt ct conqu0tes clu nroulin a e)u,"/Tnnales Ll'hisroirelcorutntique et ytciula, VII (1935), 538-563. -f) 'l-achniques i.g. his "l.c nroulini cau. Llncrivolution tr'chntqucrncdicvale." t,t civilisations,Vol. lll, No. I (19s4),pp. l-15. 1 "[.g. his "Powcr." in Charles Singcr cl a/. (eds.). A Histor.v oJ Tt,r'hnolLtgt,ll (Orford, 1956),589-622. 4";s5 plus ancicnneslncntions dc moulins hydrauliques industricls et tlc rnoulins I vcnt," Bulletin pltilologique et historiquc (1960), pp. 567-626. "l.es moulins de Toulousc au l\[o.r,en,4gc(Paris. 1953). 6"An -l'hirtcenth Industrial Rcvolution of thc ('cntury," Etonornit' Ilistory Rtvictt,,XI (194 l), 39-60. 7 Machina Ex Dco; r!'ssa-ysil the D),nantisnt of llestern (\tlntre (Canrbridge, l\lass., l96tl), passinl. R "llftdieyal Tet'hnolog.t'and Social (hange (Oxford. 1962), pp. 79-85. 9"Th. Erp"nrion of Technology 500-1500," in Carlo M. ('ipolla. cd., Tht, l'ontono Lcottorttic History, oJ Lurope, YoI. l, The Middle Ages (CollinsiFontana, 1972), pp. 155-157. 10My thrnk, to thc editors o( Scientific ,'Tnteritanfor a frcc hand hcrr'; sonre of'the idees tlevelopcrlbelow are broached in nty "A Medicval Technological Ro'olution," acquired by thcm for publication but yet to appear. tl "For evidt'nct'of a probable noria in sr'cond-(entury B.C. Egypt, sec William L. Wcstermanand CasperJ. Kraemer, Grcek Popyri in the Lihrarl'of Cornell IJniversit.t'(New York, 1926), pp.4243. I'he first unmistakable example of lhe device is that dcscribed in the late first century B.C. by Vitruvius, De architectura, Bk. X, ch.4. ed. and trans-Frank Granger(London, l914),11,304. The EnigrnaticWater-Mill t7l

t tl'1.g. in tlre works of Philo of Byzantiunt (htc second ccntllry B.('.); scc Le lirrt, (1.,sappor(ils p1(ynntulucs (,1 dcs rrnchirtt'sh.r'Jrauliques par Philon dc B.t':anca,ed. frorn thc Arab vcrsion of Orford and ('onstantrnoDlc and trans. lc Baron ('arra dc VaLrx(Paris, 1902), pp. 177-1g3. '-ln ('raLra. Anrltohtgia Ilk. lX, No.418, cd. and trans.W. R. Paton(Lr>ndon, 1925). Iil.231-23-1. l4Thc tl:rtcs for thc cpigrarr and l'or thc birth and dcath ot Antipatcr urc unknown. but dctails f'ronr his adult life arc rccordcd for thc ycars l2-9 li.C.; sce Pauly's Rcal-Iinc.t'cloltitticdt,r tlossischenAltt'rtunrsvisst'nstholt, cd. (icorg Wissou'a.I.2 (Stuttgart. 1894), col.25 14. For thc opinion that this nrav have trecn Arrtipaterof Sidon. ivho rvasborn in Tyrc:-rnd livcd in Sidon during thc scconclct'nlurv B.('., scr'Shrrucl Avilsrrr.Orr t,lrc llisror.r ol the I:.tploitatiort ol llatt,r Pou'er in l.'n,t: lsrael (lel-Aviv. 19601, pp. V[-Vll: see Plulv-Wissou'r. I. 2. rol. 2513. l<'"The r1>pcaranccof thc rotary grindstonein thc Grceo-Ronran uorld bl thc scconclcentury I).('. is firrnly, establislrcd,as it is in China by' tlrat datc; for tletailcddiscussions o{' its ;rossiblcorigin and l)attcrnsof tiillirrion. rl'ltich rctttain obscurr'. scc L. A. \1r u contlLrit lcadingfrorn lllc \tr('anl to tlrc tlcsircdhcight und sincc atrl'tlistinction is rlrcly possitrlclrottt tlre lristoriceltlata. rny rr'l'ercnce.; to r)\'('rsllotrnills rrrav [rc takcn to inclutlc troth ty pcs. ,,[-udivig[icck.Dlt,()cschiclttcdcll:'isclt.sitt|t,c'ltttischtrttttkuItllrgesl

1884t. p. 581. rnd A. W. Prrsorrs."A Ronran Watcr-l\{illin thr- Athcnirn Agora," Ht.sTteria,V, I { l9l6). 81, n. l. IR "'Avitstrr. htc. tit.. Rithurd IJcnncttand Jolrn [:lton. //i.rri)/t'ol (ltrn iI'lillirtc.ll (l.ontlon. 1t399).6-9. rncl 1.. ( ccil ('rrrwrrr."Thc Protrlcnrof l'.arlyWatcr-Mills," ,lntiquit.t'. XVIII (1944). 13.1.lnd his "A Vcrtical Watcr-MillNt:tr Safonika."Antiquit.r', XIX ( 194-5),2ll-2.12. (('urwcn includcsin his arsunrcntthe obscrvationthat llre crotr'ing cockr sisnala mral secnc.in *hrch wc woultl r-\pcct to l'tnd thc snrall,sirnplc horizontal rnill ralhcr than thc nrorc colrrPlcrvcrticll lrrrll:u cock. howcvcr,hrd bccn kno*n to crorv.jrrstorrlsidc the palaceol'thc lliglr I'ricst of .lcrusalcnr.(Mrtthew 26:69-75). tq'-[rorber, t)p. (it.. P. sq], A. P. Ushor.A Ilistoq' o] lllct'hanicolInycntions, rcv. cdn. (lloston, 1959), p. 165. antl L. A. ll{oritz,:rp. t'it.. pp. 132-l.13. '"(ittryrolth_t'.,n ljk. Xll..lr.556. ed.:rrrdtrans. IL I-..loncs(l.ondon, l92tj). V, 428-429. 21Uln.lt. o1t. tit., p. -539,and ('urucn. "Thc Problenrof I..arlyWatcr-l\lills." p- 13.1. --l:or Strabo'sdates, sce Paull-Wissowa,2nd scr.. IV, I (Stuttgart.l93l). cols.7.5.8l-85,90. -'Attct)rpts to asslgnspccil'ic clraratlcristics hcre rcrnainincorrclrrsivt: on bchalf ol'thc horizontal rnill. sec Ilcnncll and I:lton. tot. cit., anci C'ttru'cn. lot. t'it.; thc case lbr thc vcrtital nrill is arlvancedby Parsons.

'rArcl Stccnsbcrg,l.-arns and ltlills irt Dcnnurk Drrring Tvto Tltousatul Iears (('openhagen, 1952), pp.294'298. Irorbcs, np. tit., p. 594, and Whitc, Metlicyal Techrtolog.t,,p. 81, are two reccnt writers who havc indicated their gcncrll acccptancc of thc .lutluntl nrills, ar rcportetl. I dward M. F:rhy, "A llorizontal Mill at il{ashanaglass. ( ount1- Cork''' .Journalrtf tlte Cork l!istrtritol antl Archacologital &tcietT', XI (1956)' 47. says ol' tlrc Jutland find that " in thc abscncc of structural cvidcncc for a nrill on that site. the cxcavator's elainr tnust be trcatcd witlr rcscrvc":this. I bclicvc,is ir failurc to scc thc strucluralcvidencc inrplicit in thc pattern of silt dcpositsso carc'full1'de'scribcdbv thc crcavators. 'o{)1t.)a tit., Vol. IV. Pt. ). pp. 369-380,390-396. Scc also llerbort Chatlcy, "The Devclopmentof Mccltanistnsin Ancicnt China," Transactksrtsol thc Nt'*'t'otrtL,nSocict.t', XXII (194 l-42), l1'7-137. his "l:ng1'ncs," f"ngintt'rirry, ('l.XIl (1946). 388, and Whitc. llcdict'ol T't'chnolog.r',p. 81. nn. 2--s. ,o "Nccdhanr, o1t.cit., Vol. IV, Pt. 2. pp 363-165, -"lr.g1r) <'. Rcintll. "Die Intwicklung der Wasscrkraftnutzungund der \l'asserkral'tmaschincn,"]lasserkralt Jahrbuch (1924), p. 3, who statcs.but tvilhout arlcrluatcdocurncntation, thal tltc arl of using water-powerwas a Chincse invcntion that nrovcd toward Rornc along tradc routcs and through busincsstransactions. ''t]loeh,1l op. (it.. p. 540. I:orbcs.op. cit., pp. 593-597,and Bcrtrand(lillc, "l\'lachincs."in ('harlcsSingcr et al., c

40cill., "[r lnoulin ii eau," pp. 5 and 14, n.26 offers a crediblo description ol a watcr-mill by analysisof severallater medievaldocuments of negotiationsrelating to mill constructionand maintenanccdating from 1297,1370,1388, 1408 and 1468. 4lll.g. tl.T. Bossert and W. F, Storck, eds., Dos mittelalterlic,heHousbuclt trac'hdorr origirrair itt Basitze des Fiirstctt rrttt lloltlhurg-llolft'.gg-rrtoldst't'(ca.147.5) (Lcrpzir. l9l2). pl. l8: Olaus Maunus. IIist()ri0dt,gantibus st'lttcnlri

-'1.d.(l l:lcanor Knott (DLrblin.1936), p. 24. 11.794-795 "Mol nruilandcona sciathaib7 a chendraig7 a irmtiud (a rnill shaft $'ith paddlc-whccl.fastcnings and gudge'on)";for dating. sec [.ucius(iwynn, "'fht'Rcccnsions of thc Saga Togail Bruidrtc Da Dcrga." Zcits

65llorn, "Tht. Plan of St. (;all Onginal or Copy,";ra.ssillr.suggests that thc plan rs rvc have rt js l copy fronr an earlier ('arolingianprototypc. This is not to dcny, horvcvcr.that thc prcscnt plan rvasprcparcd ca. 820 for (lozbertus of St. (;all;cf. Ilorn. "On tht Author, ctc.." p. 104,and Rcinhardt,/oc. cit. 66Urkrnrla,rhuch dcr .,lbtei Sartct-Gallen.cd. Ilcrmann Wartmann (Ziirich, ttt6,l), Vol. I, p. 27. no. 24: "nrolcndinis" (7 59). A1 ""'ljornrulac Srngallcnsermiscelhncac," ch. XI. in M.O.ll.,1.r1'cs,Sec. V (llrnnovcr. lU86). p. 385: for drtc, scc ibid,. p. 319. 68,,.-"1 hrvc bten privilcgcd to sharc in some corrcspondcncebctwccn ProfcssorsLynn White and Walter IIorn, tl(' I-trs Angcfcs and UC Berkclcy respectivcly.in which thc lattor has supported thc view that thc St. (;^ll pila( dcpict watcr-powercd poundcrs uscd for tlrc Lrrcparationof bccr nrash as rvell as, perhaps, for the preparation of the cereal dish rrus, whrch was an itrlportant dietarv itent at the tirne. 69R"t'ut'il 'fessicr d(s oc!estlt' Charlas Il, lt, t'hauvc, roi tlt' Iirant'c, c11.(icorges (Paris. 1952). ll, 2l sce ( harlos l)u( angc, Ok;ssarirtrrrrrtctliae L't inlinne Iatindtis. cd. I copold Iravrc (Paris. 1937-18). ,trt. rurrtho 3 (-hrasscria). 1(l'"Brtttit'r. t)!). (it . Pl). b0l-6031 (iertrgesIispinas, Lo I ia urbaint' Lle Douai eu nt().r'en6gc (Paris, l9l3), ll, 394- 396: l\'lina Martens, "lntroduction i I'ctude rlcs rnoulins a cuu dc Ilrurt'lles." La l'olklore Braboncon. No. 149 (March. l96l), pp. I4-20; Thcodor Bitlerauf, ctl.. Dic T'ratlitionentlcs Ilochsti.ftcs Fraising (Munich, 1905). Vol. ll, pp. 547-548, no. 1772', (.trkurnlenbut'ltdcr StaLltLiibt:tk. cd. Vcrcin ltir liibeckischeGoschichte, Vol. I {Ltibcck, 1843). pp. 247-248, No. C('LXlXl Wojcicclr Wasiutynski. Ilt,gal nlvnnl' x,irednktu,it't':rt'r,nrltroxit' polskim (Warsau, I936t. pp. l4-15. '11 ' 'Op. cit.. pp. -s90-594.621-626. 12op. ,it. '"Thc tt-nn hdt0torio sccrnsto occur nrcrrcoftt-n in rclcrcntc to hcnrp nrills, i.c. hydraulic poundcrs for brcaking up tht- woody tissucsot'retted and dried stalksol'henrp and flar, tlrc l'rhcrsof whiclr werc usctl for thc rnanulacturing of linensand cordagc, Thc cristenceof thc hcnrp-ntill,horvcvcr. antl thc possibleconfusion of terms do not alter our prcscnt fbcus, for fulling-nrillsand lrcmp-rnillswcrc closcll rclatcd. e.g. scc ( arlo C'i1.rolla,il gntppo dt'i diplomi adelaidi iti Javorc dell'abbazia di I'ittcrolo (Pinerolo, l8ll9). no. 8. p. 347 and no. 9, p. 350 (1078): "molendinis, batenderiis,fullatoriis"; scc also Bautier, ibid., p. 573, and Th6resc Sclalert, Le Haut-Dauphineau Moven Age (Paris, 1926), p. li)3. T4Bouticr. op, cit., p. 572. n. 2. '"lhi

82li.g. th. "duo ntolcndina <1ucvulgo hutte dicuntur in fluvio Illstrccie" in N'[oraviain I ]69; se'c(brlc,t tlipktnraticust't t'ltistolaristrforarioc, Vol. lV (1268-1293)'cd. Antonttts Eoczek (olmiitz' 1tt45),p. 37. 83 lrl,,,ru,,rr,r,oBoit'a. Yol. XXXVI. I>t. I. eil. ,\i,atlcrtria\cicn(irrunl boica (Nlunich. I852). pp. 157. -5ll-512. Therc is rcasonto bclievethat thcsc tcnlls $cr0 also rrscdin that rcgion in refcrcnceto the cntirc opcration, that ts, bcllorvs rs wcll rs poundcrs; see I:. l\1. Rcs'. (irsrlrrclitc urtd v'irtsclta.ltlichcBctlt'ututtg dcr Oherp.ltli:ischen f;is<,rtitrtlttstricrttn tlot Atrlo'necnhis :ur /.1'it dts .l()'Ja/rrt.qcnlt ria.qaslRcgcnsbtirg, 19.50).p. 48. 84Dia Tratlitiont,n Llcs Ilot'hstiJrcsRcgcttshurg uttd tlt'sA/os/crs S. f.rtttncrortt(Munich, 1943), no. 290, 1tp. 2J2- 2ll. and no. l9ti. pp.236-237: "Snridinrulni"l1'or thc dating, scc ibrr./.,p. 173, and Iirb Nrkolaus."Allcrsburg rrr dcr OlrerpfaIt.," l'crhattdluttgcttdt's historist'hctt I'crcits ton Ohc'rpful: und Ragenshurg,X ( 1846), 100. 85 I:rk,,,,rl"uhuchdas ller:or:tunts St(ier,nar[.cd..l . Z:rhn (Crat.,187.51.Vol. l, no. l7l, p. 170. 86whi,". lllt,tlitral Tt'cltrttsktg-r'.pp. I I 2- I l -.1. 87B.rtr"nd Gille, "La came ct sa dddouverte." Techniqueset civilisations,III (1954). 8-9; Ushcr, op. cit., p. 140; Winite, Medieral Technologl', p. 79, n. 3. 88Or. ,ir., fol. l0r (vcrtical-beamstarTrp; sce F. M. Feldhaus,Die Technik der l'orzeil dt'r g,'scltichtlichenZeit und der Naturr,1lker, i\{unich, 1965, col. 915, fig. 600), and fol. l7v (rccumbcnt trip-hamrrier;sec licldhaus, op. cit., col. 1077, tig. 717). o'Ncedham,ao op. cit., Vot. IV, Pt. 2, pp. 393-395, argues that the water-powcred trip-hammer "was as characteristically Chinese as the verticel pestle stamp-rnill was liuropcan"l but his discussiondoes not resolvc the mattcr of predominance in Europeen industry during the ntedicval period.