One of the projects I’ve set for myself this summer is weeding serials with no online equivalents from the open stacks for offsite storage. I’ve done a lot of weeding over the past decade, but usually with books. Serials present harder choices. Firestone, our main library for the humanities and social sciences, is packed, and for the last decade we’ve had to ship out an item for every item brought in. And while there might be some withdrawals, especially of duplicate items, most of the weeding is for the ReCAP facility we share with Columbia and the NYPL. It’s a grueling process, but oddly enough one of the activities in which it’s necessary to theorize about if you’re going to do it well.

For books, it’s relatively easy. I factor in the age of the book, how long (if ever) since it has circulated, and its relative importance based on my own knowledge of either the author or subject in question and the language in which it’s written. Regardless of the complexities of languages and relative importance of given authors, books are always easier to deal with. By the time they make it to ReCAP, every book is completely cataloged and identified. This isn’t always true of books in open stacks. Yesterday, for example, I ran across a book that had been bound about 50 years ago with the wrong title and author on the spine. It wasn’t in our system, and according to WorldCat only three libraries have copies. Stuff happens. Anyway, since it’s cataloged and identifiable, it can always be retrieved and if necessary returned to the main stacks. If I make a mistake sending a book our two out, I can have them sent right back if anyone complains, not that anyone does. But with serials it’s a completely different story. If I send a hundred years of Rivista di Filosofia offsite, it’s probably not coming back, because I wouldn’t have anywhere to put it.

For the purposes of weeding for offsite storage, we divide the print serial titles into two types, those with online equivalents and those without, or, in the parlance of the database we use to manage this process, “B range with SFX targets” and “B range without SFX targets,” with SFX being the popular link resolver. Last year I went through 650-odd titles “with SFX targets,” and realized that while SFX is pretty good, it’s only as good as the data it gets, and it gets data from a lot of sources, not all of which are reliable. I encountered dozens of instances where we had more online that SFX indicated, and a few where we had less. Every title had to be checked. When I do projects like this, I document everything from my rationale to individual problems. I could post my nine-page, 2000-word report documenting that process, because that would be some exciting reading.

The biggest questions for me were, did we really have this online and is it from a stable source? The first was tedious but conceptually easy to answer, the second required some thought. What is a stable source? I first went the opposite route and identified unstable sources–aggregators like ProQuest, for example. ProQuest is great, but they just license their content, and some of their online journals go away. I considered JSTOR, MUSE, and anything direct from the publisher to be relatively stable, and shipped out most of those titles. (Plea to JSTOR: digitize more non-English language titles! I have a dozen suggestions to start with if you’re interested.) I also discovered that publisher backfiles for some otherwise expensive journals were pretty cheap, so I filled in some online gaps.

Some of you who have sent your JSTOR titles out years ago (i.e., some of my colleagues) might ask what took me so long with this. Mostly it was about practical space planning. Partly it’s because I really don’t like sending serials offsite, even ones we have online. That attitude is changing slowly as the technology improves, but it’s still true that there are some serials for which the online equivalent isn’t really equivalent at all. It’s slow and clunky, and browsing the print is easier and faster. I ran across a couple of French examples recently that were so clunky I yearned to have the printed journal in my hand. But space is tight, and I have close to a thousand print-only serial titles sitting on the shelves that are potential candidates for offsite storage.

“Space is tight” has to compete with another policy to save the time of the reader. What, if possible, is the best arrangement for scholars using the collection, not easiest for librarians? Sending out titles en masse is easy for librarians, but not always best for library users. Deciding that is difficult, and required (by me at least) a title by title decision in many cases. Deciding that for print-only serials, many of which aren’t indexed anywhere or at least anywhere you’d think to look, is the toughest of the weeding decisions. You can’t go by circulation data and you can’t go by age.

Even going by whether something is indexed is questionable. Okay, the Philosopher’s Index covers the last 30 years of this 60-year-old title. If the last thirty years were online, I might send out the online and keep the print in the stacks. However, of the first 250 titles I’ve dealt with, only three had any extensive indexing, and never for the complete run of the journal. And is indexing enough? Can it be browsed? An indexed journal can be “browsed” through the index, but it’s sometimes clunky. We subscribe to the major indexes in my area from Ebsco, and they do a pretty good job, but it’s hard to replicate the experience of browsing print journals. Regardless, the bulk of the titles reviewed so far have been non-English titles that standard subject indexes haven’t covered.

Some people might think that just because it’s old or a dead serial, it can go, but that’s not true, either. Sometimes, readers really need large print runs, especially if there is no online equivalent. Cultural historians might find a lot of interesting material in looking at 40 years of Life magazine, but if their best access was through an index or offsite storage then the project becomes much more difficult. Every search, every article request takes a bit more time. Most people might just want to access a particular article, but that’s not the only way people use journals.

Access to historical periodicals is sometimes crucial for a research project. For example, while working in the stacks I ran into a scholar working with a number of English, French, and German philosophy journals from 1910-1940. He’s tracing the influence and growth of  phenomenology through Thomistic and neo-scholastic philosophers in the early twentieth century, and the only way to do it is to slog through decades of journals. These aren’t online. They’re not indexed anywhere. But they are sitting on the shelves in Firestone. He spends his days walking back and forth from the stacks to the desk he’s using to work, reading a bit, finding a trace, grabbing another journal volume, and so on. Imagine how difficult that work would be if it all had to be done by recalling journals from offsite, or worse, strictly via ILL. It would be almost impossible, and extremely time consuming. Technically, it’s possible to view materials at ReCAP, but  there would still be the issue of reduced hours and access, as well as the problems of recalling 30 years of eight different journals to sit in the reading room. (I’m not even sure if that’s doable, but I assume it is. It’s a pretty slick operation.) The thing about a large research collection is that you never know how someone will be using that collection, or what scholarly projects might happen sixty years down the line that are possible only because of your wise decisions now. And if you’re thinking “who cares about such projects,” the answer is easy. The scholars who work on them care, and that’s who the library is supposed to serve. Scholarly needs should determine the collections and their accessibility as much as possible.

So far, I’m still working out the rationale. Partly, I’m making assumptions. For example, I’m assuming that no one will serendipitously discover the existence of a historical journal relevant to their research just by browsing the stacks. If you’re looking for a journal volume from 1920, for example, you probably either know what journal you’re interested in (like the scholar mentioned above) or you have a citation found in some source. This could be used to justify sending just about everything offsite, including the non-indexed content, and if I’m still doing this in twenty years and need the space, then it just might. However, there is also an assumption about what will save the time of the reader. What about the scholars who need long runs of journals, especially multiple journals? I’ve run into two instances of that this year alone, and the history of scholarship in various fields isn’t an especially unusual topic. Using imagination and sympathy, I put myself into the place of those scholars (plus I asked some). If I were using the collection for research, what would be a minor inconvenience, and what would derail my project?

A minor inconvenience might be recalling just a few volumes from offsite storage, whereas having to recall multiple volumes of multiple serials and finding a place to store and work with them would be very difficult and time-consuming. Assuming the first, short runs (10 or fewer volumes) of dead print journals can safely go. We have a lot of those from the 19th and 20th centuries. Recalling the entire six-volume run of a Russian philosophy journal and handling it is easy. Recalling decades of one or more serials and handling them isn’t. Thus, for now, long runs stay, especially long runs of live journals. Another assumption is that long runs are some indication of the relative (perhaps historical) importance of the journal. If a journal published for seventy years, it obviously had an audience of some kind. It’s historically important, and historians might be interested. Length is merely one measure, though, since there are, for example, numerous little magazines from the Modernist era that had short runs but significant cultural importance (Blast Magazine would be a good example). But in the history of scholarship, long runs are a good sign. Look how many journals in JSTOR run back into the 19th century.

I’m still working out the kinks, and there are still exceptions. Nevertheless, trying to balance the need for space with the possibility of supporting certain kinds of research is one of the trickier practical problems I’ve encountered. The complexities are partly the result of having a lot of stuff, which is what scholars want us to have. Purely patron driven acquisitions or completely online collections are only possible if you ignore the vast majority of material published before this century and outside this country. I suspect this will change over the next 50 years, but I have no way of knowing the rate of digitization of non-American materials, the intellectual property rights that might be invoked, and the affordability of that material for American research libraries. I’m trying to save space and the time of the reader. And if it turns out I’m too cautious, some successor can happily toss out the whole collection.