University of Virginia Library

Search this document 


collapse section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 
expand section 

expand section 


Page 7


When the University of Cambridge set up its printing house in 1696, nearly all the records relating to the erection of the building, and then subsequently to its operation throughout the next decade, were preserved. The annual press accounts show clearly the kinds of expenditure involved in running a small printing house; and the workmen's vouchers for composition, correction and presswork, together with the joiners', smiths', glaziers', plumbers', typefounders', fellmongers', carriers', and printers'-suppliers' bills, reveal the week-by-week operations of a printing house in a detail which is, I believe, unparalleled. In addition the Minute Book of the curators of the Press provides direct evidence of printing charges and edition sizes over the period.[10] With such a wealth of primary documentary material it has been a simple matter to construct detailed production charts for the books printed, showing their progress sheet by sheet and recording the exact division of work between different compositors, correctors and pressmen. It has been a simple matter also to offer definitive details on the wages earned and the actual amount of work done by compositors and press-crews, and to construct work-flow diagrams illustrating the disposition and organization of work within the printing house as a whole on all books and ephemera in production over any one period. It must suffice for the moment simply to observe that the patterns which emerge seem to me to be of such an unpredictable complexity, even for such a small printing shop, that no amount of inference from what we think of as bibliographical evidence could ever have led to their reconstruction. To this Cambridge evidence we may now add the invaluable record of work done in the Bowyers' printing house over the years 1730-9. The ledger which records this work is in the possession of the Grolier Club of New York and is to be edited for the London Bibliographical Society by Mr Keith Maslen. The details which it gives of compositors' and pressmen's work, sometimes week by week, also permit the accurate reconstruction of working conditions, whether for one book or for the printing house as a whole. As Mr Maslen has remarked, "Work patterns are more complex at the bigger and busier London house, but in broad outline the picture is unchanged."[11]


Page 8

Let me now look briefly at our ruling assumptions about the amount of work that a compositor and pressman might get through in an hour, day, or week. On these points, however carefully we qualify it, the evidence such as we have leads us to suppose a maximum setting rate of something like 1000 ens or letters an hour by one man and a printing rate of 250 impressions an hour at full press.[12] Taking each of these figures as averages which allow for imposition, proofing, correction, make-ready, washing and distribution, we may then, to estimate daily production, multiply by twelve to get as totals 12,000 ens and 3000 impressions. Finally, for weekly totals, we multiply by six to get figures of 72,000 ens and 18,000 impressions. These totals, we allow, are probably too high, but translating them cautiously into terms of actual book production, we might say that a quarto of five to six sheets, each containing some 10,000 to 12,000 ens and printed in an edition of 1200 to 1500 copies, would take about a week to produce if only one compositor and one full press were at work on it. Given the conditions mentioned, the logic is impeccable. Nor is it, as a method, foolish simply because we cannot know of occasional human aberrations from these norms. Yet I cannot forbear a quotation from George Eliot's Daniel Deronda:

Men may dream in demonstrations, and cut out an illusory world in the shape of axioms, definitions, and propositions, with a final exclusion of fact signed Q.E.D. No formulas for thinking will save us mortals from mistake in our imperfect apprehension of the matter to be thought about . . . [and] the unemotional intellect may carry us into a mathematical dreamland where nothing is but what is not.

The Cambridge and Bowyer papers (and they are not the only ones) make it quite clear that wages and therefore output, since the men were on piece-rates, varied considerably as between one man and another. It is not just a matter of occasional lapses or minor disparities to be cautiously conceded; it is a fundamental fact that should radically alter our whole conception of 'norms'. Moreover, any one man's income and therefore his actual output might fluctuate greatly week by week. Taking the Cambridge compositors first, one of them (John Délié) averaged 13s.5d. a week over one period of 59 weeks, and only 9s.9d. over a further period of 80 weeks. Yet on 11 May 1700 he was


Page 9
paid at the rate of £1.6s.8d. a week for setting in two weeks eight sheets of a book which required a daily average of 10,240 ens plus marginal notes. William Bertram averaged 10s.8d. over one period of 78 weeks and 10s. over another period of 47 weeks. Clement Knell and John Muckeus were content to average just over 13s. a week for long periods, and William Great 10s.3d. The only compositor to show sustained application at a high level was Thomas Pokins. His average weekly income over a period of two full years was £1.1s.5d. Yet in the five weeks up to 6 June 1702 he set some 318,000 ens, giving a daily average of 10,600, and his reward for this work was just on £1.7s.11d. a week. To reinforce the implications of these comments on the fluctuations in wages earned, we have only to look at the amount of work actually done in ens per day. Pokins's averages throughout 1702 were 6,307 (not 10,000 or 12,000) ens a day or 37,842 (not 72,000) a week. The next best daily averages for any lengthy period were those of Bertram with 5,700 ens and Knell with 5,603. Often the daily totals were well below these figures, and all of them of course are well below even such an elastic hypothetical norm as 10,000 to 12,000 ens a day. A glance at appendix II (d) below will show a similar variation of performance within the Bowyers' shop. The wages for the two-week period range from as much as £3.15s.od. to as little as 8s.2d. And this personal variation as between compositors may be paralleled by comparable fluctuations in the amount of work done at various periods throughout the year. Strahan's ledgers, although they do not yield the same detail, also show great fluctuations in the total wages he paid out week by week (B. M. Add. MS. 48,801). And the records of the Plantin-Moretus Museum, the Oxford University Press, and even the "Case-Book" of John Wilson, printer in Kilmarnock in 1803, tell the same story.[13]


Page 10

Turning now to presswork, we have precisely the same situation. One of the better performances at full press in Cambridge was that of Jonathan Cotton and Robert Ponder during the week ending 24 February 1700 when for £1.4s.6d. each they worked off 10,350 sheets on four different books, averaging well over 3,400 impressions a day. But such figures and therefore such high incomes were quite exceptional. Albert Coldenhoff, working by himself at half press, earned on average 13s.3d. a week over a period of 67 weeks, and other average weekly wages earned by various workmen at half press were, in round figures, 11s., 12s., 15s. and 18s. At full press the average receipts show the same kind of range. If instead of wage bills we again look at the actual amount of work done, we can point to Thomas Green's single-handed production of 8,250 impressions one week, and only 4,750 the next; or to the work of Ponder and John Quinny who printed the following numbers of impressions week by week from mid-June to mid-July 1700: 15,200; 13,800; 9,700; 12,700; 10,700; 17,000 and so on. The hypothetical norm, you may recall, was 18,000.

Well, conclusions? Simply that our hypothetical figures are too high? Certainly that seems to be true, but the implications are I think more far-reaching. One is that we have perhaps failed in our historical sense, too readily imputing our own twentieth-century ideas and interests and the assumptions of our own society — especially our economic assumptions — to men whose attitudes to work were quite different from ours. We cannot afford to disregard contemporary social conditions and pre-Industrial Revolution attitudes. One careers adviser pointed out in 1747 that many pressmen played a great part of the time, and Benjamin Franklin took much the same jaundiced view of the British workman. As one economic historian puts it, "the conditions of life and habits of the people were all against the monotony of regular employment;" and again: "Contemporary evidence indicates that few cared to take advantage of their opportunities." The mass of labourers, said Sir William Temple, work only to relieve the present want.[14] Our society today reacts to the stimulus of high wages because modern society can satisfy a wide range of wants. But if wants don't increase, there is little point in working for anything beyond the bare necessities, and good wages will only suggest to a workman "an opportunity


Page 11
to avoid part of his toil."[15] An anonymous writer of 1728 makes the point well: "People in low life who work only for their daily bread, if they can get it by three days work in a week, will many of them make holiday the other three or set their own price on their labour" (my italics).[16] One of the reasons why Elizabethan printers tried so often to exceed their allowed number of apprentices may have been that apprentices could be commanded to work regularly where journeymen could not.[17] So although we may today think of piece-rates as an incentive, it would seem that in the 17th and 18th centuries they were the employers' best protection from men who had no intention of working any harder than necessary for food and drink. A journeyman's output, that is, was conditioned largely by what he was able or willing to earn. As Moxon revealingly says, "they are by Contract with the Master Printer paid proportionably for what they undertake to Earn every working day, be it half a Crown, two Shillings, three Shillings, four Shillings, &c." (my italics). That is, he speaks quite casually of a performance difference of one hundred per cent.[18]


Page 12

The significance of this system of individual contracts — operating as far back as 1631 and even, I would claim, 1591 — has I think been overlooked. What it meant in effect was that a workman need work no harder than he had contracted to do, and only if he fell behind his contracted figure and then kept another waiting might he have to make good his colleague's loss of work. It was the master's job to accommodate these variables, not the workmen's. That the master had to make payments and organize production schedules on the assumption that men worked at different speeds is quite consistent with the Cambridge and Bowyer evidence and any other that I have seen; nor is it, so far as I know, inconsistent with any of the classically 'demonstrable' bibliographical proofs. But its consequence — the normality of non-uniformity — is an uncomfortable one for any methodology.

It might be thought that the fluctuations in output which I have noted are — in part at least — to be expected if there were too little material to keep the men working at high capacity week after week, a state of affairs all too likely in a small provincial academic press. Yet, once more, the evidence available shows a similar pattern at other presses. In fact, both the Cambridge and Bowyer records suggest that fluctuations in the volume of work would be reflected more readily in the number of workmen employed than in the actual level of work done by any one man.[19]


Page 13

Where output varies so markedly from man to man and period to period, any reliance on 'norms' would seem to imply an almost irresponsibly large burden of probable error; polite concessions to occasional departures may serve a while as palliatives but general statements that can be so persistently falsified whenever any concrete evidence appears to test them are poor premises for advanced argument. Moreover, I have for simplicity here dealt mainly in averages; the actual figures are infinitely more varied and any attempt to trace the total complex patterns week by week, even with all the documentary evidence, is like trying to record the changing images of a kaleidoscope in the hands of a wilful child.

So far I have discussed, too, only the most elementary variables, and have left aside all question of edition sizes or the ways in which work was organized. Assumptions about the edition sizes of early books usually take account of the 1587 ordinance of the Stationers' Company which, with one or two exceptions, forbade the printing of more than 1250 or 1500 copies of any book from the same setting of type, although as McKerrow remarks "we have no certain knowledge of how long or how carefully the rule was observed."[20] And occasionally edition sizes have been related to the bibliographical evidence of skeleton formes, since it is understood "as a general principle that in any book printed on a single press two sets of headlines will appear only if the book was printed in an edition large enough for composition to keep ahead of presswork."[21] For the later 17th century and the 18th century the problem of variability in edition size is acknowledged to be more complex. Yet even in the early 1590's a pressman, Simon Stafford, giving evidence in court, pointed out that the number of sheets printed in any one day might vary considerably, reflecting different edition sizes, since "they had diverse numbers upon Severall bookes and the


Page 14
numbers did alter."[22] At Cambridge in the early 1700's sermons were printed on average in 650 copies, the minimum being 400 and the maximum 1150. But the mean, as distinct from the average, was nearer 500 copies. One book of theological controversy was printed in a first edition of 750; a second edition of 1000; a third edition of 2000; and a fourth edition of 1250. Yet another of the same author's books ran through three editions in the order of 750, 500 and 500; and another was printed in two editions of 1000 copies each. Other figures for edition sizes of books on a fairly wide range of subjects are: 350, 500, 522, 550, 700, 820, 1050, 1150, 1250, 1500 and 3000. Again the Bowyer papers show a similar variation, apparent even in the Voltaire editions cited in appendix II (g). And the same point is made by the edition sizes given in Strahan's and Ackers' ledgers, and more particularly in those I quote both for the part issues of A New General Collection of Voyages and Travels and the monthly numbers of The London Magazine. Quite apart therefore from these figures invalidating any hypothetical norm (for any particular case, given the degree of variation, would seldom correspond to the assumptions of a guiding hypothesis), the need for a master printer to juggle with such varying totals, as well as the varying abilities of his workmen, reinforces what has been merely hinted at so far about work organization.

Normally in bibliographical analysis we are concerned with a particular book, usually a work to be edited, and that great range of printing which our literary interests have not led us to examine must be largely ignored. This is as it should be for life is short and, as Professor Todd has indicated in a disarmingly amusing note, some books are more valuable than others.[23] But the consequences of such selectivity cannot be ignored if bibliography, as the study of the transmission of literary documents, is to continue to lay claim to the serious intellectual status that Greg established for it. I shall return to this point. Meantime I wish to look briefly at a very common assumption, though not a universal one, in bibliographical analysis. It is the assumption that even if the whole resources of a house were not directed towards printing the book under examination, at least one compositor and one


Page 15
press-crew would be set to work fairly consistently on it.[24] Under these conditions we might expect five to six sheets a week to be completed.

Again, however, should we not take pause? The occasional prospectus might serve to put us on our guard, for few require printing times of anything like even five sheets a week. And in fact some surviving Cambridge agreements offer delivery times of — as near as matters — one sheet a week. Out of some 36 books of ten or more sheets produced between 1698 and 1705, only 7 were printed at an average rate of more than 2 sheets a week. Suidas, the Greek lexicon, was printed at the rate of 3½ sheets a week, and the remainder of this group of 7 books progress on average at a rate of between 2 and 3 sheets a week each. For 14 books the average was between 1 and 2 sheets, and for 15 books it was no higher than a single sheet and often it was less than that. The Cambridge evidence cannot be discounted by noting that some of these books required careful correction, for the evidence from the Bowyer and Ackers ledgers points exactly the same way.[25] Nor does there seem to be any necessarily significant relationship between the total amount of work on hand and the rate of progress for any one book.

The force of these examples is simply that the Cambridge and Bowyer presses, like any other printing house today or any other printing house before them, followed the principle of concurrent production. Obviously the variety of runs gave greater flexibility in the organization of presswork and permitted more economical use of the


Page 16
several sizes of type available; but, however efficiently total production was organized, the system inevitably meant that individual books took longer to print than we might have thought likely, just as most books today take far longer than the productive capacity of the machinery would lead us to expect.

The important point, however, was made by Professor Todd many years ago. Under conditions of concurrent production, he remarked, "the book is only one of several components in a more extensive enterprise, and thus exhibits only a portion of the information necessary for its analysis. Until the other portions have been located and the various pieces reassembled in the pattern originally devised the puzzle will remain unsolvable" ("Concurrent Printing," pp. 45-6, 56). And again: "in instances of concurrent printing the bibliographer must examine all the books so related before attempting the analysis of any. To do less than this . . . is to learn little or nothing at all." These remarks cannot be repeated too often, but unfortunately Professor Todd's qualifying comments have tended to minimize their application to "the larger establishments of the eighteenth century [where] the facilities were certainly adequate for simultaneous work on several projects, involving, in some instances, independent groups of compositors and pressmen, in others, the same group intermittently employed, first on a few sheets of one book, then on a few of another." And his more immediate concern, the interpretation of press figures, whose "very presence implies unsystematic piecework engaged in conjunction with other miscellaneous endeavours," has also perhaps encouraged the belief that concurrent production may very well have been a feature of large printing houses in the 18th century but not of smaller and earlier ones.[26]

Such a view must be abandoned. No amount of historical quibbling can neutralize the plain facts of the Cambridge documents: an earlier and smaller printing house, never using more than two presses, often one and a half, and occasionally only one, habitually printed several books concurrently. So far as I am aware there is no primary evidence whatever to show that any printing house of the 16th, 17th or 18th centuries did not do likewise.


Page 17

Our mistake here I think goes back to a misreading of an observation by McKerrow. He remarked that "for a printing house to be carried on economically there must be a definite correspondence between rate of composition and the output of the machine room."[27] Notice how all-inclusive his terms are: printing house, machine room. I should like to offer now two quotations which take their origin in McKerrow, and I should like you to notice how in each case McKerrow's valid general statement is transformed into an invalid particular statement. First, Professor Turner:

In the Elizabethan printing shops, a cardinal principle of efficient operation was, we suppose, that composition and presswork should proceed at the same rate. If material could be set into type faster than the press could run it off, compositors had to waste time; conversely, if presswork went faster than composition, the pressmen would stand idle. Given pressmen and compositors of reasonable skill, the chief factor determining the speed of the presswork was the size of the edition . . .
at this point we become involved in a particular statement
. . . and the chief factor determining the speed of composition was the amount and difficulty of the text material to be set into each forme. Ideally one forme ought to be machined in the time required to distribute the immediately preceding forme and to set and impose the next, and in the case of books which got a great deal of text into each forme, as the [Shakespeare] Folio did, this ideal relationship could be approached only if two compositors could set simultaneously.[28]
Professor Hinman asks:
What plan would ensure the most satisfactory ratio between the time necessary to set one forme of the contemplated book into type and the time needed to print off the desired number of copies of such a forme? — for an efficient balance between composition and presswork was one of the prime requirements of successful printing house operations in the earlier seventeenth century. (Printing and Proof-Reading, I, 45n.)
This last is a quite different and seriously distorting assumption: that an economic relationship between composition and presswork is necessary on any one book for the business as a whole to be successful. The position is really so much more complex; indeed the more variables a printer has to juggle — in numbers of compositors and full or half press-crews, in their individual capacities, in edition sizes, in the number


Page 18
of books on hand, and in the demand for ephemera — the more chance he has of making them compatible and therefore of making his business as a whole economically successful.[29]

Such a conclusion is not inconsistent with the figures given earlier to demonstrate the variable levels of production achieved by a printing house; they simply reinforce the point that an 'economic' disposition of men and materials could only be achieved complexly — and that the fine considerations of timing implied by many studies devoted to the analysis of a single work may be a world away from the reality. Relax the time scheme ever so slightly, and a whole house of bibliographical cards comes tumbling down. In particular the correlations often traced between edition size, number of compositors, skeleton formes and presses, must look very different if translated to a context of concurrent printing. But I anticipate.

If concurrent production was much the most efficient way of running a printing house, as distinct from the most efficient method of producing a single book, how then, under these conditions, was work apportioned? What kind of range does a compositor's work, for example, show week by week? Although it was by no means rare at Cambridge for a compositor to have a monopoly on any one book, the work was usually shared. Of a sample of 118 Cambridge books, only 50, or 42%, were set by a single compositor; and if we except very short works like sermons, the percentage drops to 24. Moreover it was unusual for a compositor to work for any long period on one book to the exclusion of all others — usually he would be setting type for two or three concurrently. Of the 13 compositors whose work for Bowyer over a two-week period early in 1732 is recorded in appendix II (d), only one was engaged on one book alone.

Undoubtedly the main considerations determining the allocation of work were simply a compositor's freedom to do it and the availability of type. If a compositor had no other work on hand he would be transferred to any that might be offering and for which type was available. For normally, even when two or more compositors worked on a book, they did not work together setting sheet and sheet about. What usually happened was that one took over where the other left off and then composed as many sheets as the master found convenient or


Page 19
as other commitments allowed. A quarto edition of Virgil may be cited to make the point. Bertram set sheets A-E, then Crownfield took over and set F-3R, next Michaëlis set 3S-3Z, Bertram then resumed setting, continuing to 4F, Délié was brought in to set 4G, Crownfield took over once more with 4H and finally, after Crownfield had retired at 4O, Bertram set to 4R and finished off the book. Although four compositors were involved, and although Crownfield worked on two sections of the book and Bertram on three, on no occasion were any two men setting simultaneously. Nor incidentally would it be true to say that they were setting alternately. And it would certainly be quite wrong to assume that work was divided so that the book might be printed more quickly. Meantime of course each Cambridge compositor was also concurrently engaged on some other book or books.[30] In appendix II (g) examples 2 and 4, Bowyer's two English editions of Voltaire, show the same pattern of work. Whereas the printing house printed, and individual compositors normally set, several works concurrently, the composition of any one book would therefore usually be a simple matter of progression from sheet to sheet, or from one group of sheets to another, by consecutive compositors. It follows that the compositorial pattern within any such book will rarely offer adequate evidence in itself of the productive conditions but will have been determined by, and will reflect the exigencies of, the general pattern of work in the printing house as a whole over a period of months or even years.

Turning now to the other half of the equation — presswork — we must again affirm that the most efficient disposition of work, given the variables to be reconciled, could be achieved only by a highly flexible system. As with composition, the actual manner in which work was apportioned to the men would have depended in part on the number of men employed and the amount of work on hand. But regardless of the size of the plant it seems unlikely that a particular press consistently served the compositor or compositors setting a particular book. At Cambridge, it is quite clear, any press-crew might get any sheet of any book to print off, and consequently it was rare for any book of more than two or three sheets to be printed solely at one press. The production tables for books printed at Cambridge show quite conclusively that the use of two press-crews was a perfectly normal procedure and had nothing whatever to do with increasing the speed of production. If a forme was ready for printing, it went to whichever


Page 20
crew was ready to take it, although usually it seems that a sheet would be printed and perfected by the same crew. The position in the Bowyers' shop is much more complex, for it is clear that formes, not just sheets, might be sent to any press which happened to be free and that any sheet might well be printed at one press and perfected at another. Again the Voltaire and Baxter examples in appendix II (g) make it plain that this practice was very common.

In a large shop the distribution of work is likely to be more complex simply because there are more routes open for a book to take; but in both large and small shops there is such strong evidence of fluctuations in the number and strength of press-crews that a pattern of work must very often reflect such changes too. At Cambridge there was considerable variety in the number and composition of the press-crews, ranging from half press only during the first half of 1699, to full press only during the second half of the same year and a good part of 1700, and various combinations of half and full press during other years. Only for a brief period in 1701 were two full presses in operation simultaneously; but we find (unexpectedly, for the usual arrangement we might think would be for the two men to work together at a single press) two half presses at work during 1705, 1706 and part of 1709. There were frequent changes in the composition of the crews during the period from 1699 until early 1702. In later years the normal arrangement — if you will permit the term — was one full press and one half press. But the varying patterns make it extremely difficult to assume a norm. So too in the Bowyers' shop: 3 presses were in use between 24 Dec. 1731 and 15 Jan. 1732; 4 between 17 Jan. and 29 Jan.; 4 (but only 3½ crews) between 31 Jan. and 12 Feb.; 5 between 14 Feb. and 26 Feb.; and 6 (but possibly only 4 full and 2 half crews) between 28 Feb. and 18 Mar. (Just to complicate matters further for the analyst, press no. 7 was in use throughout the whole of this period and the press figure 7 appears on some of the sheets which it printed, but at no time were there as many as seven presses in use.)

So a press-crew, just like a compositor only more so, would usually be working on several books at a time, with "All work to be taken in Turn, as brought to the Press, except in such Work as may require Dispatch, or the Compositor will want the Letter . . .".[31] The simplest way of using the crews most efficiently was not to try to maintain a strict relationship between a particular compositor and a particular press — the varying edition sizes and varying output of the men would


Page 21
have made this very difficult — but, given the presswork which was offering, to apportion it so that each crew always had something to go on with. This was the easiest way of accommodating the varying runs required for the different books in production at the same time. In thinking otherwise we may also have under-estimated the flexibility of the common press itself as a machine. Each press had several friskets, ready cut to common formats. It was a very simple matter to change them over; and since the sheets were printed wet on a sopping tympan, the type bit deep into the paper and careful make-ready was unnecessary — certainly it did not need the care of the modern kiss-impression. Technically there was no reason therefore why the press should not work to a number of compositors setting several different books, perhaps within one day, certainly within one week.

It is more than time now for us to re-unite the two halves of McKerrow's equation — the 'rate of composition' and 'the output of the machine room'. We have seen that, both at Cambridge and in the Bowyers' shop, books were produced concurrently. This meant not only that several books were in production at the same time but that each workman, whether at press or case, was often engaged on several books more or less at once. If we are correctly to reconstruct the detailed operations of a printing house — even a very small one — or a true account of the printing of any one book, we must therefore do it in a way that shows the complete pattern of work in its full complexity.

The diagrams given in appendix I (a) and (b) are an attempt to do this. They show precisely how all work on hand at Cambridge was allocated between 26 Dec. 1701 and 28 Feb. 1702 and convey some impression of the flow of work.[32] Again I am tempted to quote George Eliot and say that the sheets seem to follow only what she calls "the play of inward stimulus that sends one hither and thither in a network of possible paths." Suidas, an exceptional case, demanded the almost undivided attention of four compositors working in pairs on each volume; and Crownfield, Bertram and Pokins spread their work over two to three books each. When we look at the distribution of work to the different press-crews, we note that with one exception every compositor or pair of compositors sent work to both presses; and, moreover, that the work composed during these weeks was in many cases


Page 22
printed by any of four different press-crews. If we take a single book — Psyche, either volume of Suidas, or Whiston's Short View — the point to be made is much the same, that the various sheets of any particular book were likely as not printed at more than one press. The details given for the Bowyer shop in appendix II (a) to (e) cover two distinct periods: first, from 26 Dec. 1730 to 6 Feb. 1731, and second, from 31 Jan. to 26 Feb. 1732. Their testimony in witnessing to the disposition of work is consistent with that of the Cambridge records but the relationships revealed between the several productive units, whether compositors or pressmen, are very much more complex. The information given reveals, for example, the number of pages set by each compositor, the edition size, and the number of formes printed by each press; in the case of appendix II (a) we can also see the peculiar arrangements between presses 1, 3 and 7 for printing and reprinting the Defence of the Present Administration. I take this evidence to be quite conclusive. It shows that the essential procedures for the distribution of work were the same for a larger and later shop as they were for an earlier printing house with only two (or more commonly one and a half) active presses. It shows that although we should doubtless be right to assume — allowing for certain social attitudes and conditions we have mentioned — that composition and presswork as a whole were fairly economically balanced, it would be quite wrong to conclude that this balance was either necessary or possible for work on any individual book.