Monday, October 21, 2013

A database for the Hoyle copyright

I'll return shortly with the sequel to "What was the Hoyle Copyright Worth? (part one)." Perhaps this essay will make clear the reasons for the delay.

In "Researching Copyright" I discussed the many tools I used for learning who owned shares in the Hoyle copyright. They were the Stationers book of registry, imprints, publisher's records, bookseller trade sales, receipts, and newspaper advertisements. I have many examples of all of these for Hoyle, including more than 500 newspaper advertisements for 18th and 19th century Hoyles. I tend to have PDF files of the various sources reasonably well organized on my computer. To compile and analyze the data, I rely on dozens of Word documents and Excel files. As the data have become more numerous, it has been harder and harder to keep the Word and Excel files in sync.

I have been contemplating putting everything in a database for a long time and finally started a couple of days ago. I built a database to manage the copyright data: books, booksellers, imprints, advertisements, and more. For the technically minded, I built the database in sqlite3 and use python where extra processing is required. It took about 25 hours to get something useful and I'm pretty pleased with what I can do. Here are a few examples of questions I can now easily answer:

From the 1740s until the 1860s, many booksellers bought and sold pieces of the Hoyle copyright. Which booksellers held pieces for the longest time?

bookseller  from_date   to_date     years     
----------  ----------  ----------  ----------
Baldwin     1755-12-24  1835-05-27  80        
Longman     1800-05-15  1868-07-09  68        
Lowndes     1771-11-12  1821-01-02  50        
Wilkie      1767-12-12  1814-01-11  47        
Newbery     1771-11-12  1800-05-15  29        
Crowder     1757-12-22  1785-12-08  28        
Mawman      1800-05-15  1826-03-26  26        
Bladon      1771-11-12  1796-03-05  25        
Payne       1779-11-13  1804-05-12  25        
Scatcherd   1796-03-05  1821-01-02  25        
Stewart     1796-03-05  1820-02-18  24        
Osborne     1745-10-26  1767-12-12  22        
Law         1775-06-09  1796-03-05  21

The Baldwin firm comes out on top, ahead of Longman, Lowndes, and Wilkie.

Of course these were not individuals, but families or firms who held the copyright for the better part of a century. One can see the evolving names in imprints and advertisements:

from_date   to_date     first_name    last_name     suffix                             
----------  ----------  ------------  ------------  -----------------------------------
1757-01-01  1767-12-12  Richard       Baldwin                                          
1771-11-12  1813-12-24  R.            Baldwin                                          
1820-02-01  1826-03-26                Baldwin       Cradock, and Joy                   
1835-05-27  1835-05-27                Baldwin       and Cradock                        
1800-05-15  1803-08-03                Longman       and Rees                           
1808-05-24  1808-05-24                Longman       Hurst, Rees and Orme               
1813-12-24  1820-02-01                Longman       Hurst, Rees, Orme and Brown        
1826-03-26  1826-03-26                Longman       Hurst, Rees, Orme, and Co.         
1835-05-27  1835-05-27                Longman       Rees, and Co.                      
1842-03-02  1842-03-02                Longman       Brown, & Co.                       
1847-03-03  1868-07-09                Longman       and Co.                            
1771-11-12  1779-11-13  T.            Lowndes                                          
1785-12-08  1820-02-01  W.            Lowndes                                          
1771-11-12  1779-11-13  J.            Wilkie                                           
1785-12-08  1796-03-05  G. and T.     Wilkie                                           
1800-05-15  1803-08-03  G.            Wilkie                                           
1808-05-24  1813-12-24                Wilkie        and Robinson                                

It would be possible to research the history of these booksellers in the British Book Trade Index to see if what I'm seeing for the Hoyles accurately reflects deaths, and successions.

I've written elsewhere about the 1774 case of Donaldson v. Beckett, eliminating the common law perpetual copyright in England. Who owned a share of the Hoyle copyright before that decision?

Booksellers with a share in the Hoyle copyright

For a number of reasons this report was hard to produce--it took some help with python. Note the disposition of Thomas Osborne's share with his death in 1767 and the proliferation of owners shortly thereafter. The report is even more interesting when it is extended in time, but that would be hard to display here.

Another question: Which book stayed in print the longest? I looked for books which were advertised the longest after publication date. The results are preliminary, as I've entered only a subset of advertisements in the database, but even the early results are interesting:

book                  publish_dt  advert_dt   years     
--------------------  ----------  ----------  ----------
1745 Laws of Whist    1745-10-26  1751-11-12  6         
1750 Osborne 10       1749-10-21  1755-12-24  6         
1761 Chess            1760-12-30  1766-01-08  6         
1761 Chances          1760-12-24  1764-01-18  4         
1800 Jones Direction  1800-05-15  1804-05-12  4         
Remarkably, most of the Hoyles were in print three years or less. The exceptions are worth more research and more discussion.

I identified The Laws of Whist and Directions for Breeding Game Cocks as poor sellers from another source, catalogues from bookseller trade sales. In the essay "The (missing) Laws of Whist Designed for Framing" I noted that the Osborne sale of 1767 offered 325 copies of the Laws more than twenty years after it was published. See the discussion below for further evidence of the slow sales of the Laws. Similarly in "Hoyle's Games Improved, Charles Jones (1800)" I noted there that Directions was a poor seller, with bookseller Wilkie's remaining stock going unsold at an 1814 bookseller's trade sale. It is comforting to note that trade sale catalogues and newspaper advertisements tell the same story.
[Aside: In my research I have focused on the trade sales primarily for sales of the Hoyle copyright, and have not searched exhaustively for the much more frequent sales of books unless copyrights were offered at the same sale. I'm only beginning to appreciate how much could be learned from this unimaginably time-consuming effort. That work would disclose more examples of poor sellers and could also help estimate print runs. For an example, consider the "eleventh" edition of Hoyle's Games. Six months after it was published, 350 copies were offered at the Hodges trade sale. As Hodges had owned a one-third share, the print run was likely 1250 or 1500 copies. See The Hoyle Copyright in Hoyle's lifetime."]
The database helps another way: in the essay on Hoyle's Games Improved, I speculated that the price for Directions might have been a shilling or two. In fact, advertisements shows it sold for sixpence, something that if I noticed before, I had not recorded on the right spreadsheet.

The appearance of the "10th" edition of Hoyle's Games on the list does not tell the full story. In fact the "10th edition" is a reissue of the "8th" edition dating back to 1748. See "Reissues of Mr. Hoyle's Treatises (1748-1755)." I have not yet done the work to connect multiple issues when they are the same edition (and indeed, it can be difficult to tell which issue is being advertised). I've often wondered whether Osborne overestimated the demand when he had the "8th" edition printed, or whether there was standing type and multiple impressions were made.

Interestingly, Chess and Doctrine of Chances do not appear to have been great sellers. For more on the latter, see this essay.

A last example: What books were advertised at more than one price? Here, I would expect to find situations where the booksellers were forced to lower prices.

book                  CNT                 
--------------------  -----
1745 Laws of Whist    2                   
1751 Laws of Brag     2                   
1757 Osborne 11       2                   

Three books where advertised at multiple prices. Checking the specific advertisements, I find:

date        paper                 book                  s.   d.
----------  --------------------  --------------------  ---  ---
1745-10-26  London Evening Post   1745 Laws of Whist    1    0      
1746-01-14  London Evening Post   1745 Laws of Whist    1    0      
1747-11-07  London Evening Post   1745 Laws of Whist    1    0      
1748-03-05  Whitehall Evening Pos 1745 Laws of Whist    0    6      
1748-04-30  London Evening Post   1745 Laws of Whist    0    6      
1751-11-12  London Evening Post   1745 Laws of Whist    0    6      
1751-01-22  General Advertiser    1751 Laws of Brag     2    6      
1751-01-25  General Advertiser    1751 Laws of Brag     2    6      
1751-02-28  Whitehall Evening Pos 1751 Laws of Brag     1    0      
1756-12-21  London Evening Post   1757 Osborne 11       3    0      
1757-06-10  Public Advertiser     1757 Osborne 11       3    0      
1757-12-22  Public Advertiser     1757 Osborne 11       3    6      
1757-12-24  Public Advertiser     1757 Osborne 11       3    0      
1757-12-27  Public Advertiser     1757 Osborne 11       3    0      
1760-01-03  Public Advertiser     1757 Osborne 11       3    0         

As I observed earlier, the Laws of Whist did not sell well, and Osborne lowered the price from a shilling to sixpence in 1748. Jolliffe had the same problem with the Laws of Brag, lowering the price from two shillings sixpence to a shilling almost immediately. Brag itself likely had the same problems, but I only have inferential evidence of its price. As far as the "eleventh" edition of Hoyle's Games, apparently the printer made an error in setting the December 22 advertisement.

I have a lot more data entry to do, primarily advertisements and trade sale data. Once I do that, I'll be ready to do a better job of  part 2 of "What was the Hoyle copyright worth?"

Well, I'm enjoying my new toy. What other questions should I be asking?

Saturday, October 5, 2013

What was the Hoyle copyright worth? (part 1)

Readers of this blog know that I am obsessed with the Hoyle copyright. I see Hoyle as a brand, rather than an author; a business endeavor, rather than literature (see "Continuities and Disruptions"). Understanding who owned the copyright (see "Researching Copyright"), what it was worth, and how it was exploited is, I believe, the key to understand the Hoyle canon.

One way to assess the value of the copyright is look at what booksellers actually paid for it. For example see "The Hoyle Copyright in Hoyle's Lifetime" where we see transactions for shares of the Hoyle copyright implying a value of £187 10s. in 1763 and £261 in 1767.

But how might a bookseller decide how much to pay for a copyright? In this post I am going to present a simplified model for valuing the Hoyle copyright. If the results seem reasonable, I will do the extra work of removing the simplifications.

Timing of Valuation

The model will value the copyright only at the book is about to be reprinted, the time time when it is most valuable. Booksellers make money from the copyright by printing new books for the cost of printing, paper, advertising, etc., and selling the books to retail booksellers (and perhaps wholesale booksellers as well; see the discussion below). When a book has just printed, the opportunity to produce new books and thus to make money from copyright ownership must wait until it goes out of print, reducing the value of the copyright. See the discussion in Belanger, Booksellers' Sales of Copyright: Aspects of the London Book Trade 1718-1768. Ph. D. Dissertation. Columbia University. 1970 at pages 105-8. When we compare the model's prediction of copyright value with actual sales, we will have to note the timing of the sales with respect to reprinting.


The Longman Archives give us detailed information about print runs and expenses for the Charles Jones edition of Hoyle's Games Improved and the G. H. edition of Hoyle's Games Improved and Enlarged from 1796 to 1868. I am going to use the Longman data from 1796 to 1826 because the data are more consistent year-to-year, letting me work with averages without distorting things greatly.

Printing of the Jones Hoyle. Data from the Longman archives.
The data show that the Jones Hoyle was reprinted every five years with production expenses averaging £304 and retail sales £1082.
Aside: There are some details I'm overlooking. For example the publishers would often issue sections of the book separately. See for example "Hoyle's Games Improved, Charles Jones (1800)" where I note that the publishers excerpted the treatise on game cocks and issued 500 copies. That would account for a small portion of the expenses (500 of 42,500 sheets) and bring in small additional revenue (although it was a very poor seller, with perhaps half the stock unsold fifteen years later). Similarly, portions of the book were issued separately with each edition from 1803 to 1820. It would be possible to extend the model to account for these separate issues which were governed by the same copyright.

The model values the copyright at two points in time, five years apart. These are both times when the book is about to be reprinted, so we won't have the timing problem discussed above. The major assumption of my model is this: the value of the copyright is the same at the beginning and end of the five-year period. This assumption would be silly for a typical book where the sales are uncertain. If the book sold well, the value of the copyright would increase; if it sold poorly, the value would decrease, perhaps to zero. Hoyle is an unusual case, a perenniel best-seller, where the booksellers could count on selling an entire print run of 3000 or 4000 books every five years.It seems reasonable that with unchanged prospects for the book, the value of the copyright would be unchanged as well. Perhaps the same model would apply to other best sellers such as bibles, almanacs, and school books.

A lesser assumption is that the bookseller will sell 1/5 of the books at the end of each of the five years. This is unrealistic because the book would likely sell best when it was newly published, with diminishing sales over time. Second it defers each year's sales to the end of the year--a monthly model with decaying sales would be more realistic.

Finally, I am ignoring the generous credit terms that prevailed at the time. Publishers would not pay printers immediately--six months credit was not uncommon. Similarly, publishers extended generous credit to the wholesale and retail booksellers.

The Model

The model assumes the following cash flows:
  • At the end of year 0, the bookseller purchases the whole copyright. 
  • At the end of year 0, the bookseller pays for a print run (£304)
  • At the end of each year 1 through 5, the bookseller sells 1/5 of the books at some percentage of the retail price of £1082 (see discussion below).
  • At the end of year 5 the bookseller sells the copyright for amount originally paid. 
In the model I will vary two items. First is the discount rate. In a cash flow model, future cash flows must be discounted to a present value. I've looked for historic interest rates in England in the early 19th century and they seemed to cluster around 4 or 5%. I'll look at discount rates from 3 to 6%

Second, it is not appropriate to credit the copyright owner with the full retail price of the book. Some of that profit is attributable to retail bookselling rather than publishing. Indeed there was a third role of wholesale bookselling that existed at the time. There would be separate prices from the publisher to wholesaler, wholesaler to retailer, and retailer to the public. Even where a bookseller played all three roles, we must determine a publisher's price to determine how much revenue is attributable to publishing rather than distribution.

From my reading in the early 19th century book trade, it appears that the wholesale price is about 60-75% of retail. See for example James J. Barnes, Free Trade in Books. A Study of the London Book Trade since 1800. Oxford: Clarendon Press. 1964. p22. I haven't seen discussion of a publisher's price, but there is evidence that for Hoyle, it was 40% of retail. When Hoyle was reprinted in 1803, 72 copies remained unsold from the 1800 edition of 3000 copies. The copyright owners include the value of those unsold books as an expense against printing the 1803 edition at 2s. per copy. The retail price was 5s., providing evidence of a publisher's price of 40% of retail. (from the Longman Archive). For the model, I'll look at the publisher's price varying from 40 to 60%.

The Model's Output

Given a discount rate, and a publisher's price, and given that the value of the copyright is the same in the end of year 0 and year 5, you can uniquely solve for the value of the copyright.

Value of Copyright
You can see that the model is not terribly sensitive to the discount rate, but is hugely sensitive to what percent of the retail book price is ascribed to the copyright owner versus the distributors. 

I plan to share the model with friends in finance and with book historians. Next essay, I'll share their feedback and compare the modeled value of the copyright with what booksellers actually paid. More soon!

Friday, October 4, 2013

Serendipity at the Library

In my essay "A Research Trip to Cleveland," I wrote about two Hoyles I saw in the White Collection that were gathered in nines. I quoted Fredson Bowers from his Principles of Bibliographical Description:
There do exist, however, a very few extraordinary books for which it would be acceptable to use odd index numbers when the odd leaves indicate a consistent method of printing a whole book and not simply of an isolated gathering...Jacob Blanck in a recent article on Washington Irving's Salmagundi pamphlets (1807-1808), which often exhibit the initial gathering in 9's and even in 11's, refers to several early nineteenth-centruy books regularly gathered in 9's. (page 228-9)
Today I went to the San Francisco Public Library to read the Blanck article, "Salmagundi and its Publishers" in Papers of the Bibliographical Society of America, Volume 41, First Quarter 1947, pp1-32.

Salmagundi is a periodical written by Washington Irving, his brother William, and James Kirke Paulding. It was issued in 20 parts in 1807 and 1808. The Blanck article describes each part according to the (pre-Bowers) bibliographical standards of the day and provides an interesting overview of the publisher, David Longworth of New York. Blanck details a number of Longworth's eccentricities, such as puffing his own business in city directories he published in the early 19th century and habitually not capitalizing "new-york."

As to the nine-leaf gatherings, Black notes:
Another of Longworth's eccentricities becomes evident when one collates Salmagundi and discovers the existence of many gatherings in nine, and one in eleven, the full significance of which can be appreciated only by a printer.6a 
The footnote continues:
6aBooks in nine are so infrequent that other examples may be of interest. Longworth was responsible for at least one other: Willia Dunlap's Ribbemont, Or The Feudal Baron, A Tragedy in Five Acts..., New-York. Printed And Published by D. Longworth...1803; this collates: [A]-D9. Longworth had no monopoly and this peculiar for of book-making. Another example is Joe Miller's Jests...A New Edition, London: Printed [by T. Kaygill, Strand] For Duncombe...[n. d., ca, 1810]. The "Jest Book," a bibliographical jest indeed, collates: [A]1, B-H9, I8.
And the serendipity? The publisher David Longworth is quite familiar to me. He published the second edition of Hoyle in America, a book I describe in "The New Pocket Hoyle, New York, 1803." The Blanck article made me wonder whether Longworth's New Pocket Hoyle might be gathered in nines--I had never collated it.

I came home to examine my copy and found it was gathered in eights, with a prosaic collation of A-G8. I must say, I was a bit disappointed! But it was amusing to seek out the PBSA article for a footnote and be entertained by the entire article!