For Printed Materials: The Probability of Appearance
- By Bruce McKinney
Joel Munsell: the fastidious bibliographer
By Bruce McKinney
It is becoming possible to estimate the print runs of perhaps 150,000 antiquarian items, whose printing histories were never recorded, based on formulas that compare detailed printing records provided by Joel Munsell to the number of copies found in the OCLC, appearances at auction and sightings on listing sites and eBay. The goal is a new statistical measure that projects "probability of appearance" for books and ephemera. The internet has made it possible for millions of items to be offered but what is the likelihood what you are looking for will show up? I think we can know.
Munsell's material, published over a fifty year period in the 19th century, covered an extensive range of subjects and was printed in formats that range from 8-page pamphlets to 400-page books. He documented his printing history and it comes down to us today as a basis for comparison to copies in institutional, auction and listing venues. He published a directory of 2,268 printings and documented a thousand items, published between 1834 and 1870, with bibliographic details and the quantities printed.
For book dealers, collectors, auction houses and libraries such information may in time become part of a printing's essential facts. The initial print run is but one factor in calculating "probability of appearance" but it is an essential one that combined with appearances in various databases, historical data and bibliographic details may make it possible to construct an index with implications for all collectible printed materials.
The theory is simple algebra and the objective to solve for the unknown. Munsell provides half the variables and modern databases and listing sites the other. In Munsell's case, with both sides known, we should be able to establish the range of variation with some precision and apply this range broadly to material that lacks the quantity printed but is otherwise bibliographically complete.
For this we owe a debt to Joel Munsell, the Albany, New York printer, who maintained extensive records of the various books and pamphlets he printed during his career. In Bibliotheca Munselliana: A Catalogue of the Books and Pamphlets issued from the Press of Joel Munsell from the year 1828 to 1870, he detailed 2,268 book and pamphlet printing jobs including the quantity printed for about 1,000 of them. Using the information he provides it's possible to compare his quantities with their appearances in library inventories, at auction and on listing sites. With these numbers we can estimate survival statistics and in time "probabilities of appearance."
In this article I examine Munsell's production for the period 1834 through 1840 for which he identifies 104 items. I then focus on 64, all books and pamphlets, for which he provides the quantities printed and compare these quantities to the number of copies found in the OCLC today. Let's take a look.
|
For Printed Materials: The Probability of Appearance
- By Bruce McKinney
Munsell's first book
The aggregate print runs of these 64 items total 78,270 and average 1,223. The question next is how many of these specific items and how many copies of each are found in libraries, the assumed safest and most logical place for such material to be found. To find out I searched for each in the OCLC, the Online Computer Library Center database of books et al held by about 40,000 participating libraries. For 34 of the 64 items I find at least one example in the OCLC. For the other 30 no examples are identified. For the 34 with at least one copy a total of 324 copies are found. Therefore, for about half [47%] of Munsell's early production, there are no examples in the OCLC and for the other half an average of nine. For 14 of these items 10 or more copies are found representing 255 of the 324 examples identified in the OCLC. For the other 20 items there are only 69 copies. There is the distinct impression of feast and famine.
Munsell Material: 1834-40. An Analysis by Presence and
Absence of Copies in the OCLC
| Publications & Copies |
Present |
Absent |
Total * |
OCLC Copies |
| 400 or less Copies |
6 |
16 |
22 / 4,445 |
11 |
| 401 to 1000 |
17 |
13 |
30 / 22,375 |
190 |
| 1,001 to 4,000 |
9 |
0 |
9 / 19,450 |
74 |
| 4000+ |
2 |
1 |
3 / 32,000 |
49 |
* Number of Munsell print jobs and total copies printed
The size of the print run and the number of pages provide some clues about survival rates. For the 64 items in Munselliana with specified print runs the total of all copies printed is 78,270 and the 34 items for which at least one example is present 53,585 or 68.5% of the total. Thirty items representing 31.5% that were printed in an aggregate quantity of 24,685 are not found in the OCLC. It's apparent some items survive in disproportionate numbers and others perish disproportionately. For the items that are found in the OCLC their holdings comprise 6/10ths of 1% of the original print runs and the average print run is 1,576. For the 30 items for which no examples are located in the OCLC the average print run is smaller: 823 but this doesn't explain why some items are held by many libraries and other items have not a single copy. Not surprisingly the number of copies printed is a complicated variable involving judgments, preferences and luck or lack of it.
|
For Printed Materials: The Probability of Appearance
- By Bruce McKinney
An obtainable early imprint
Munsell's 1834-40 Printings by Number
of Pages and Size of Print Run
| Print runs/Number of pages |
1 to 16 Pages |
17 to 71 Pages |
72 + Pages |
| 400 or less Copies |
6 |
11 |
0 |
| 401 to 1000 |
6 |
15 |
4 |
| 1,001 to 4,000 |
3 |
4 |
3 |
| 4000+ |
1 |
1 |
1 |
Let's look at statistics comparing the number of pages to the number of copies located in the OCLC. For works of at least 72 pages the survival rate in the OCLC is 1 in 184. Thus if a 1,000 copies were printed and the item had at least 72 pages we can estimate there will be 5 or 6 copies in the OCLC. For material, mostly pamphlets, with 17 to 71 pages the survival rate is 1 in 221, for pamphlets of 16 pages or less 1 in 502. When few copies were printed and the item had 16 pages or less the likelihood of survival is very small. When the print run as well as the number of pages is small we see mass extinction.
Munsell's 1834-40 Printings in the OCLC Analyzed by Number of Pages and Size
of Print Run, Divided into Present and Not Present and Expressed as Percentages
| Publications & Copies |
1 to 16 Pages |
17 to 71 Pages |
72 + Pages |
| |
present |
not present |
present |
not present |
present |
not present |
| 400 or less Copies |
33% |
67% |
36% |
64% |
|
|
| 401 to 1000 |
28% |
72% |
63% |
37% |
80% |
20% |
| 1,001 to 4,000 |
100% |
0% |
100% |
0%
| 100% |
0% |
| 4000+ |
0% |
100%
| 0% |
100% |
|
|
Both small print run and small page count are strong nagatives for survival. Such material is automatically uncommon although not necessarily important or valuable. Most examples are simply curiosities.
|
For Printed Materials: The Probability of Appearance
- By Bruce McKinney
Bad news was good business.
Now let's look for these items on the listing sites. I primarily used ABE but also searched via AddAll. For the 64 items in this study I found 3 examples, two of them for the same item and AddAll found no examples that weren't also on Abe. For the other 62 printings I did not find even one copy. This suggests the ratio of appearances between the OCLC and listings sites is 324:3 but that is not true because the OCLC is a fixed database while the listing sites are variable. Said another way when an item is posted to the OCLC it is probably there for life while when it is posted to a listing site it is there only until it sells. In the past 18 months I have purchased 14 examples of 10 of the 64 items which I found on listing sites and eBay.
By the end of the year I expect to complete an analysis by decade. I'm particularly interested to see how time affects the number of appearances. Content is another factor and becomes important as Munsell's career advances and his judgment about content, marketing and presentation improve. For the 1840s, 1850s and 1860s we have 900 more records and in the fullness of time we will both better understand Munsell's production and I hope, by extrapolation, develop formulas by which to predict frequency of appearance and its inverse: rarity.
For the final decade of Munsell's career, the 1870s, we have only his printed output as Munselliana effectively ends in 1870. For his final decade I expect to propose a theory to predict printed quantity where none has been provided and from there it is a short step to predict probability of appearance for virtually all old material. If I can do this I know Joel will be very pleased. In his own life he never quite reached the stars. Perhaps one hundred and twenty-six years after he died he'll take this final step.
To examine a complete list of all materical used in this study Click here.
If there are others who are interested to participate in this project I would appreciate hearing from you. In time the Munsell study must be simply one source of perspective. It will take a meta-study of other documented printers to confirm or amend the projection theory as it emerges from this first study.
The analysis is engrossing and I believe the application to older material of all types a distinct if yet today unproven possibility.
Reach me by email at bmckinney@americanaexchange.com
|