Chapter II

Publishing and Library Costs: The Book

Word count: 4500

∣📄 p.3 The point of reference from which the problem of reproducing research materials must be examined is the system by which publishers produce books and libraries acquire, arrange, and store them for use. The financial policies of institutions, the stack room and reading room equipment of libraries, the training of library personnel, as well as the research habits of scholars, are standardized in favor of the use of books of several hundred pages, from 6″ to 11″ in height, costing the purchaser from $.005 to $.02 a page.

There are other forms in which re- search material is made accessible, but in any of these other forms there is created a special problem of finance or administration. Maps, newspapers, “oversize” books, manuscript collections, films, collections of pamphlets and broadsides present special difficulties and cause special expenses regardless of their acquisition costs.

Scholarly books published in America are sold at an average price of $1.20 per 100 pages, scholarly periodicals at $.70 per 100 pages.¹ A library dollar buys 73 pages, approximately 32,000 words, in book form: or 150 pages, 67,000 words, in journal form. If research materials in book ∥ form can be sold unbound at the price level of periodicals or with good library binding at the normal book price, they can be fed into the library system of the country without requiring significant changes in present acquisition policies.

Analysis of Production Costs²

The cost accountancy of book production is very complicated, but the basic fact is simple: the unit cost of producing a book falls rapidly as the size of the edition increases. This is the essential accountancy difference between the printed and the photographic (photostat or film-slide) copy of a text.) The unit costs in the one case diminish rapidly as the size of the edition increases; in the other case they are more nearly constant for any number of copies. When the prospective market, and hence the size of the salable edition, is so small that the unit cost of the volume rises above the level at which the normal purchasing policies of institutions or individuals will float the enterprise and accomplish adequate distribution, the point is reached where commercial and scholarly needs conflict with each other. The critical test of book costs, for the purposes of this survey, is the minimum size of an ∣📄 p.4 edition which can be sold within the normal price range.

Another striking characteristic of book costs is the high percentage of the sale price which goes to pay for other things than manufacturing costs. One university press reports that 33% to 50% of all money received goes to pay for distribution. Another reports that it calculates that only 40% of the cost of a book is the actual production cost. Two very complete estimates and analyses of book costs are available in the surveys made by Mr. O. H. Cheney and Mr. Donald Bean.³ Mr. Cheney reports on commercial publishing, Mr. Bean on scholarly publishing. Both writers calculate the disposition of the money actually received by the publisher. In the case of commercial publishing this is much less than the amount paid for the books by the consumer, because the retail dealer’s profit of 40% must be included. The commercial publisher’s dollar is the consumer’s $.60.

The dealer’s discount in scholarly publishing is reckoned at 30% by Mr. Bean, and the calculation is further made that the scholarly publisher is subsidized by authors and institutions to the extent of 35% of his costs. Therefore the publisher’s dollar in scholarly publishing is the consumer’s $1.07. For every dollar paid by the consumer for a commercial book, the publisher gets $.60; for every dollar paid retail for a scholarly book, the publisher gets $1.07. ∥

These relationships between cost and price must receive further analysis, but in the meantime note must be made of the distribution which publishers make of each dollar⁴ they receive. (See Table I.)

Part of the larger percentage of manufacturing costs found in Mr. Bean’s figures may be due to the inclusion of some items which Mr. Cheney classes as overhead. Moreover, Mr. Bean’s whole system of accounting seems to disregard separate books after the first year of sales and to total revenues from later sales to pay overhead costs. If an 18% overhead, presumably from sales of old books, is figured into Mr. Bean’s costs of scholarly publishing, the result is:

The larger manufacturing cost ratio in scholarly publishing is the result of the smaller edition.

If every edition were invariably sold out, the retail price of a book could be fixed by dividing the cost of the whole edition into the number of copies and adding a profit. Actually, because of the risk element, the selling price must be much ∣📄 p.5 higher. It is usually fixed at three or four times the manufacturing cost. Dr. Arthur Meiner in the German Boersenblatt proposes a formula[^n5] for determining selling price, which is illustrated with typical figures by Mr. Bean:

The largest element of print cost—the composition or typesetting —is a cost per page which is independent of the number of copies printed. The composition of a 400-page book, for example, at $2.25 per page is $900.00. If the number of books printed from this type is 1000, the cost of composition is $.90 per copy; if 10,000 are printed, the cost is $.09 per copy. While other elements of printing cost do not decrease so markedly as the quantity increases, the same principle applies to a lim∥ ited extent. Thus the remaining costs of paper, presswork, and binding may be $400.00 for 1000 copies or $.40 per copy, and $.31 per copy or $3100 in quantities of 10,000 copies: The total manufacturing cost per copy of this imaginary book is, therefore, $1.30 in lots of 1000 copies, and $.40 per copy in quantities of 10,000 copies. Stereotype or electrotype plates costing approximately $1.25 per page additional have not been included, although a necessary cost if provision is to be made for additional impressions.⁵

The formula suggested by Dr. Meiner for pricing this hypothetical book runs as follows:

The percentages which fill in this formula in ordinary commercial publishing are estimated by Bean as follows:

Consequently, the book described above, with manufacturing cost of $1.30 in lots of 1000, $.40 in lots of 10,000, would be priced:

But if the manufacturing cost is $.40, it would be \({$.40 \over .25}=\$1.60\ (\$.004\ per\ page)\).

The writer has learned that one great university press uses the following method of pricing, which bears some resemblance to the Meiner formula:

These figures indicate that even when an edition of 1000 is sold, the income from sales at the normal price of $.022 per page will not pay the cost of the edition. But the usual sale of a scholarly book is less than 1000 copies.

Mr. Marshall and Mr. Bean made independent studies of the normal size of an edition of a book written by a scholar for scholars. Mr. Bean thinks that the average first-year sale of a research book is between 500 and 600 copies; the average first-year sale of books published by university presses between 350 to 600 copies; and the ultimate sale from 900 to 1000. Mr. Marshall, after studying the histories of fifty-nine titles published by learned societies between 1925 and 1929, computes⁶ the following averages:

These small sales certainly could not be expected to support publication at a price per page so nearly normal. Naturally, the publishing is not self-supporting. In fact, the figures show that the expenses of manufacture and administration involved in the publication of these fifty-nine titles ran to $115,000, while the net proceeds from sales to apply against these expenses came only to $41,000 in round num∥ bers. The loss was, therefore, $74,000 on fifty-nine titles, or $1254 per book. Sixty-four per cent of the money came to these publishing projects as a subsidy. This is the ratio for publication by the learned societies themselves.

The costs and returns from a typical research book are shown in Table II on the following page.

It is to be expected that the learned societies would select for publication the titles least likely to have a wide sale. The general average for scholarly publication is computed by Mr. Bean to stand on a somewhat different level. He calculates that the proportion of publication cost met by subsidy is 35%.

It can be concluded that under present publishing practices, a sale of 600 copies will bring in 66% of the cost of publication; a sale of 200 copies will bring in 33% of the cost of publication; and that no book can be expected to get the publisher out of the red until after sales have passed well beyond the 1000 mark.

Mr. Cheney’s estimates of average sales in commercial publishing (which includes some scholarly publishing) confirm these conclusions. Mr. Cheney finds that 58% of all fiction titles sell over 3000 copies, and 19% of all non-fiction, titles sell over this amount. Of fiction titles the following distribution is normal:

9 books in 10 sell over 1,000
8 books in 10 sell over 1,500
7 books in 10 sell over 2,000
6 books in 10 sell over 2,500
5 books in 10 sell over 4,000
4 books in 10 sell over 5,000
2 books in 10 sell over 10,000
1 book in 10 sells over 20,000 ∣📄 p.7

Thus, everything points to the marked difference between scholarly and commercial publishing economy, because of the lack of a wide market for the tools and products of research scholarship.

The Volume of Scholarly Publishing

Mr. Bean has gathered figures which make it possible to put the quantity of scholarly book and periodical publishing in its financial perspective. The number of different printed books in existence has been estimated at twenty-four million titles. Mr. Richardson of the Library of Congress thinks there are eight million of these in American libraries. Every year approximately 150,000 new titles appear, about half of which possess some academic interest. Ten thousand of the 150,000 new books are published in America, and of these possibly one-half possess an element of interest to the scholar. But the number of works of pure research scholarship (including materials and tools of research) is much less than this. A count of the American books accessioned by the Library ∥ of Congress in 1927 shows that in that year only 1119 books were published for research scholars. This count excluded the many popularization textbooks and books of applied science.

There are about 20,000 periodicals, including newspapers, issued in the United States, and of these, according to Mr. Bean’s calculations, 291 are important research journals issued by scholars for scholars. The output of the total publishing business in America, according to the census of manufacturing, runs to two and a half billion dollars a year. Book and pamphlet publishing is conducted at a cost of two hundred million dollars. But the cost of publishing these 1119 research books and 291 research journals is estimated by Mr. Bean at approximately six million dollars a year—three and a half million for books, and two and a half million for journals. Only 1 ⁷⁄₁₀% of the money spent for book and pamphlet publishing goes to publish research books and pamphlets. The cost of publishing for research scholars is only ¼% of the publishing budget of ∣📄 p.8 the country. It is, therefore, inevitable that the main technical and business developments in publishing should follow other needs than those of research scholarship.

Nevertheless, the scholars are in a position to control their own part of the industry, for the funds for scholarly publishing are drawn almost exclusively from the scholars themselves, being paid directly or indirectly by the educational institutions of the country. Sixty-five per cent of the money received for the sale of research journals comes from the libraries together with the members of the societies who are for the most part salaried by the educational institutions. But the money received from sales does not defray the whole cost. In scholarly book publishing, 35% of the money is received in the form of subsidies which are in some cases paid directly by the institutions, in other cases by authors who are salaried by institutions.

Here are two great enterprises: the publishing trade with an annual budget of two and one-half billion dollars, and the colleges and universities with an annual budget of five or six hundred million. Their interests overlap in the small item of six million dollars’ worth of scholarly ∥ research publishing. Organized scholarship can, if it will, bring this area completely under its own control. He who pays the piper can call the tune.

The Scholarly Market

There are no adequate statistics on library expenditures for research books, but a few figures are available to indicate the scale of library purchasing operations. The United States Office of Education reports that the total annual expenditure for books, pamphlets, and periodicals reported by public, school, college, and other 1ibraries (totaling 5803 libraries) in 1929 was sixteen million dollars, of which almost fourteen million was spent for books published in America.⁷ In 1931 library book-purchasing power was nineteen million dollars.⁸ Most of this money is undoubtedly spent on books of little or no scholarly interest. Mr. James Thayer Gerould has kept records of the sums spent for acquisitions by forty of the larger university and college libraries. His figures, rounded out to thousands of dollars, stand as in Table III for the year 1930–31.

These forty libraries spent a total of $3,226,000 and added to their resources, a total of 1,134,510 volumes in 1930–31. The list does not include certain ∣📄 p.9 of the largest reference libraries in the country, such as the Library of Congress, the New York Public Library, or the John Crerar Library.

If each of these forty libraries buys every one of the research books (1119) published in a given year, and subscribes to all of the important research periodicals (291), the total amount carried on each budget for the support of American scholarly publishing is only $6350 per library, or a total of $254,000 toward the six-million-dollar publication bill.

The smaller university libraries and the medium or larger college libraries must, therefore, be reckoned as the main contributors to American scholarly publishing. Mr. Bean estimates that 975 such libraries are in the market for research books. A careful study of the budgets of about 200 of these smaller libraries has been made by Professor Randall of the University of Chicago.⁹ His findings show that 205 college libraries spend every year for books $964,000; the average amount annually devoted to book purchase is $3900; the average number of books added to the library is 1500 titles per year. Mr. Randall thinks there are 600 libraries of this class. These studies suggest the conclusion that the lesser college libraries may be spending as much as three to four million dollars a year, but that the smaller colleges are often unable to provide adequate library facilities for instruction of students, and hence have little margin to be used in helping to finance the reproduction of research material.

Aside from the college libraries, there are many agencies which in special fields may help to defray the costs of scholarly publishing with their purchasing power. Mr. Franklin F. Holbrook’s study,¹⁰ compiled under the direction of the Joint Committee on Materials for Research, lists 458 agencies of which only a small proportion are college libraries, and of which many are active in acquiring research materials. Historical societies, law libraries, and special libraries of all kinds round out the institutional market for research publications.

∥ Aside from the libraries of institutions, there is a market for scholarly publications among the scholars themselves. The Association of University Presses maintains cooperatively on addressograph plates an educational directory of 67,000 names, distributed about as follows:

These lists are used by the various university presses in circularizing announcements of books. Presumably they constitute the active market list. Fifty great university and college libraries, a thousand small ones, several hundred other institutions, and 67,000 scholars—such is the market for scholarly publishing in America. it is here that the 200 to 600 copies of the scholars’ research books find their sale.

Library Costs

The publishing cost and purchase price of the book are not the heaviest expenses that have to be met in making it available to a scholar in a library. Library finances must be given a place in the total picture, not only because the library is the agency through which the educational institutions pay 65% of the sales revenue, or 42% of the publishing bill, but also because the different methods of reproducing materials fit differently into the scheme of library finances.

The money spent by libraries for acquisitions is only a small part of the cost of maintaining the library system. The costs of putting books on the shelves, administering them, and housing them are greater than the costs of acquiring them.

The cost analysis of library administration is a disputed field, in which the files of the Library Journal report a prolonged controversy. The most active work in cost analysis has been applied to the study of cataloguing cost—the cost of ∣📄 p.10 putting the book on the shelves. The figures which seem best attested are those of the University of California—$.72 per book. Mr. Paul North Rice, a specialist in cooperative cataloguing, thinks that the real cost of putting a book on the shelves is between $1.50 and $2.00 per volume. Figures collected by the government show that university libraries spend from 50% to 72% or their funds for other than acquisition purposes. Cornell spends 50%, Yale spends 72%; the normal amount is about 66%.

Costs of administration are likely to be unusually high when the material cared for is not in normal book form. Manuscript and pamphlet collections and collections of ephemera are often made by libraries without the use of much purchase money, their cost being principally administrative.

The capital cost of housing a library is net included in’ this figure, since land and buildings are not often paid for out of current funds. The Cleveland Public Library, which pays for its main building out of revenue, finds that interest and twenty-five-year amortization on a four-million-dollar building, for which land was donated, take up 39% of the money, leaving 61% for book purchase and administration. If the cost of the land had been charged to the library, the percentage of revenue devoted to housing would have been nearer to 75%. The New York Public Library maintains an enormous investment in land and buildings. In libraries of this kind the increasing quantities of material and numbers of readers may call for the doubling of the plant every twenty years, and the housing cost may be more than double the administrative cost, or four times the cost of purchasing books. If a book costs $1.00, such a library must spend $2.00 to administer it and $4.00 to house it. A person who gives a book to a library under these conditions is really putting the library to an expense that may be six times the value of his gift. That this housing cost cannot be dismissed into the remote future but must be accepted as a normal element in library financing, is indicated in Mr. Randall’s study of ninety-four college libraries. He found that forty-eight of these will be filled in ten years, and twenty-five within a five-year period. Out of 128 library buildings, sixty-four report either ∥ that they are already beyond their working capacities so far as stack space as concerned, or that they will be beyond them within ten years.

What is a normal cost-per-book for library buildings? It is often figured very roughly at $1.00 per volume; that is to say, a million dollars will buy a building to house a million volumes. But stack room is much cheaper than the building as a whole. The president of Snead and Company, specialists in library shelving problems, estimates the following:

On the basis of present prices, library buildings can be put up at a cost of from $.409 to $.50 per cubic foot. The stack portion of a building without the stacks would probably cost from one-third to one-half of this figure. The equipment of the stack room would cost from $.30 to $.40 per cubic foot. In figuring stack room, a figure of about two and one-half volumes per cubic foot is approximately correct.

According to these figures, the cost of stack room per book would vary from $.062 to $.10, but the cost of equipment in the stack room would increase this by an amount anywhere from $.12 to $.16 per book. The total cost per volume stands, therefore, between $.18 and $.26 for stack room only. The difference between this figure and the “dollar-a-volume” estimate is accounted for by the requirements of space for administration, reading, and study.

Just as the manuscript or pamphlet collections cause an abnormally high administration cost, so newspaper collections cause an extraordinary housing charge. Shelving alone costs $1.00 per newspaper volume, and the cubic footage or stack space needed with the shelving runs this cost up to $2.00 or $3.00 per volume. This fact has an important bearing on the problem of preserving newspaper files, for it may be necessary to choose among preserving the existing paper stock, making photolithographic copies in reduced size on permanent paper, or making microcopies on film or glass. By reducing the size of the newspaper one-half, housing costs can be cut three-fourths or even more. Another point at which housing costs affect policies in ∣📄 p.11 the reproduction of research material appears in connection with the microcopy or
filmslide. To the extent that the filmslide replaces the book, the housing costs become negligible. In the Library of Congress the visitor may see copies of a million manuscripts stored in filmslide form upon a few inconspicuous shelves.

Notes