I owe many of my favorite ideas about library preservation assessment to birds and trash. I have a friend who does ornithological research into the effects of habitat development and disturbance on breeding shorebirds, specifically the changing relationship between humans and arctic shorebirds in response to a new landfill in Barrow, Alaska. Birds being generally more migratory and less long-lived than library books, his methodologies for data collection and analysis have always been more agile than mine.
It was only natural to think of him when I happened across a copy of Sokal and Rohlf‘s Introduction to Biostatistics at the Strand a while back. I snapped it up immediately, and it’s turned out to be an enjoyable way to brush up my statistics. One of the reasons that I like biological analogies is the necessity of considering change over time in biological processes. Put another way, biology and preservation are both concerned with the developments of the relationship between inherent and environmental factors. Indeed, the role of “inherent vice” is one of the earliest theoretical models for preservation, even though the term has been deprecated in favor of a permanence/durability model. Sokal and Rohlf make a very powerful observation as they introduce their ideas of biostatistics:
[19th century biologists] tried to demonstrate that biological processes were nothing but physhiochemical phenomena… Many biologists, even to this day, have retained the tradition of strictly mechanistic and deterministic concepts of thinking… In biology, most phenomena are affected by many causal factors, uncontrollable in their variation and often unidentifiable. Statistics is needed to measure such variable phenomena, to determine the error of measurement, and to ascertain the reality of minute but important differences. (Sokal and Rohlf. 2 ed. p.4)
The survey methodology that I have been working with hangs somewhere between these modes of evaluation. Some questions are addressed to inherent elements of the object in hand while other are concerned with the impact of environmental elements.
Because those environmental factors are interpretive and generally impossible to observe in action, I am trying to ensure that the data collection focuses on objective observations that support an interpretive method, but remain clearly separable from that interpretive model. For instance, we record data about the condition of the objects parts on a four-point, forced-choice scale, that generally follow this pattern:
- 3: Good Condition
- 2: Some wear or minor damage
- 1: Severe damage
- 0: Unusable
In data collection, we record the condition as observed: The binding is unusable (0), for instance. Our interpretation of that data relies on a deterministic model of how materials are affected by their environment and how that leads to the expression of their innate capabilities.
Here’s a real example, using data from surveys performed by my classes at Indiana University and the Pratt Institute. Because we work from many collections, determined by the happenstance of student interest and affiliation, randomly selecting items from those collection, we take these surveys to be a sufficiently unbiased (or biased in sufficiently meaningless ways) representation of the overall condition of books in libraries, period.
More specifically they represent the general condition of books in the greater New York-Bloomington library coverage area. Given the libraries within that zone and the vast scope of their collections and service models, I feel pretty comfortable in believing the results of these surveys to be pretty relevant to any librarian, any where.
In two recent surveys, we arrived at the following information about library bindings and hardcover publisher’s bindings:
| Case Style | Avg. Condition | Avg. Age | No. of Samples |
| Publisher’s Hardcover | 2.51 | 1978 (29 yrs) | 497 |
| Library Binding | 2.75 | 1964 (43 yrs) | 299 |
| All Styles | 2.57 | 1972 (35 yrs) | 977 |
This data is interpreted to mean that library bindings are more durable than publisher’s bindings, since they are, on average, both older and in better condition. This is to be expected, given the planning that goes into library binding (you can find more information on the current library binding standard at the website of the Library Binding Institute).
Our deterministic model of binding performance suggests that certain elements in a book’s construction will allow it to weather use and abuse more effectively. A hard case, for example, or stronger hinge-cloths and super-cloths. This is why we assume that hardcover books do better than softcover and why library bindings, with their robust hinges and super-cloths, do best of all. So let’s add some data about publisher’s softcover bindings.
| Case Style | Avg. Condition | Avg. Age | No. of Samples |
| Publisher’s Hardcover | 2.51 | 1978 (29 yrs) | 497 |
| Publisher’s Softcover | 2.50 | 1984 (23 yrs) | 145 |
| Library Binding | 2.75 | 1964 (43 yrs) | 299 |
| All Styles | 2.57 | 1972 (35 yrs) | 977 |
Softcover bindings show the same average condition as hardcover bindings, but they reach that condition a few years earlier and we have significantly less data about them. Here, our basic model of hard-cover durability isn’t completely contradicted, but it’s not strongly supported. There is an also an indication that paperback durability may be higher than the received wisdom would suggest.
There is still some investigation to be performed on other correlated data, but I hope these data are sufficiently interesting in themselves. The questions that emerge from this are quite provocative. Consider the price gap between hardcover and softcover bindings as opposed to the cost of a library binding and it’s easy to endorse preventative binding of softcover books.
I think that this points to a larger danger in preservation education and training. It’s very easy to be proscriptive: send materials for library binding, store them in acid-free or buffered enclosures, allow no food and drink, maintain a particular temperature and relative humidity.
There’s nothing inherently wrong with any of those recommendations, of course. In real library situations, however, deterministic approaches can lead to proscriptive preservation recommendations. These can overlap with one another to the point of incoherence and make implementation difficult or even impossible, and all of this before we can even respond to the many other requirements placed upon a library collection.
However, when these data are considered inside of a real library context, a proscriptive model of prevention can fall flat. It’s increasingly common for large research libraries to move their collections into off-site storage long before they reach the age at which damage of any sort becomes common. Once in off-site storage, their use is dramatically decreased and environmental conditions are improved. Hard cover versus soft cover versus library binding cease to signify.
This brings us back to the “necessity of considering change over time” that I mentioned at the outset. From this vantage point, there’s a solid case to be made for doing no preventative binding at all, allowing nature to run its course, and backfilling against the actual damage with a more comprehensive treatment program for damaged materials. Taken together, these measures could yield a net improvement in the condition of the collections while costing the library less in staff time and outright dollars spent.