In Defense of Remote Storage

By

Terry Martin

[revised October 19, 2005]

See also, Michael G. Chiorazzi, Books, Bytes, Bricks and Bodies: Thinking About Collection Use In Academic Law Libraries

When university officials first suggested that Harvard libraries should make greater use of off-site storage, I was dubious. When a graduate of the Harvard Business School, the president of Iron Mountain, Inc., a record storage company, was hired as a consultant, I became more so. When he proposed bar coding all books, shelving them by size, and placing them on three-story tall industrial shelving twenty miles outside of Boston with no possibility of browsing, I joined all my librarian colleagues in opposing the plan. Of course, the University went ahead anyway. And I am now forced to admit that the Harvard Depository - untraditional as it may be - does work.

As of July 1, 2005, the Harvard Law School Library had 668,393 volumes stored off-site, 39% of its total book stock. We have another half million volume equivalents in non-print format as well. Of the on-site collection, 100,000 volumes are in local storage and another 250,000 rare books are in compact stacks. The open stack collection amounts to only one-third of our total holdings, excluding entirely electronic resources.

The volumes in remote storage tend to be older but not rare materials, often in foreign languages. They include both serials and monographs. They are selected largely by date, although current imprints from selected foreign sub-jurisdictions like the canton Bern and superseded editions of any date are sent off-site as well.

Our on-site collections are at steady-state: for every new volume added, one must be removed. One out of three is withdrawn, sometimes replaced with microform, sometimes sent to an overseas library, sometimes discarded. The other two are sent to the Harvard Depository in Southborough. http://hul.harvard.edu/hd/

We now process for remote storage about 30,000 volumes a year, freeing space on the open stacks for new books and making small in-roads in the number of books still stored on campus in the damp basements of dormitories, inaccessible by elevator or book truck.

Processing materials for off-site storage is time-consuming and expensive, though it is work we would eventually do in any event. However, the Harvard Depository will not accept books that are not in the online catalog, not processed with item records, not bar-coded, not clean and free from mold, not well-bound or otherwise encased. So we are forced to do bibliographic and preservation work before items can be shipped to Southborough.

The cost of storage is about 40 cents per volume per year. Storage costs have been rising about 2% annually for a decade. Patrons may place orders directly through the online catalog. The items come the following day, since we do not pay the premium for expedited two-hour deliveries. The cost of retrieval and return is about $4.25 per item, less than it was ten years ago. We project about 6,300 retrievals this year, 1.6% of the total volumes stored off-site.

Let me suggest three reasons for a library with a substantial collection to consider using off-site storage. First, it's economical. Second, it's better for the books. Third, it can be easier for the readers.

An analysis done ten years ago, when we were considering the trade-offs between remote storage and additional library space on campus, concluded that if less than 6% of a collection circulated from remote storage annually, remote storage was cheaper than building new stacks in Cambridge. Local costs may vary, but few libraries face higher real estate costs than we do. It would seem we can continue to send out a substantial number of additional volumes before the retrieval rates rise and remote storage becomes uneconomical.

I should add that we have a process in place, which we occasionally implement, for bringing back permanently from remote storage books in demand.

Since many of the volumes we send to HD are getting on in years, are fragile, or would be hard to find if we needed to replace them, the remote storage retirement home is clearly better for them as objects than the jostling and shifting they would encounter in the metropolis of our open stacks.

Finally, having these little-used volumes off the open shelves is actually easier for our students. The books they are usually looking for are not hidden on a shelf filled with these old fogies. But they can order an old fogey right through the OPAC and pick it up tomorrow. Our remote storage facility is now implementing a document delivery service that will send PDF copies of articles either to a library printer or directly to a student's workstation.

Is something lost with remote storage? Of course. Shelf browsing has to be replaced by online bibliographic browsing by call number, which is a cumbersome technology as yet. Serendipity is lost, though one might argue serendipity is only necessary because librarians haven't indexed their collections very well or researchers never learned to use indexes very well. Citation indexing does make up for a lot of that, though not for historians mining primary sources. And, of course, librarians cannot impress touring visitors quite as much with the vast extent of their collection if much is out-of-town.

Library collections have become much more mixed format affairs anyway. Viewed in one light, Harvard has only one-third of its collection on the open stacks in book form. But as I tell any complainers, “Let me know when you've read those and I'll replace them.”

In fact, if we were to apply the 80/20 rule, perhaps we should store even more. Some of the general information laws and certainly the laws of bibliometrics seem to point in that direction.

The Pareto Principle - named after Wilfredo Pareto , a nineteenth century economist and sociologist - states that a small number of causes is responsible for a large percentage of the effect: usually a 20-percent to 80-percent ratio. Bradford's Law - the concept that the most significant articles to any given field of investigation are found within a relatively small group of journal publications - is simply a specific application of this general observation. Eugene Garfield, the founder of ISI, expanded on this bibliometric heuristic even further. Garfield's law of concentration argues that there exists a large overlap between disciplines and that material only tangentially related to a particular discipline often comprises the core of another discipline.

In other words, if you know your collection well, you can afford to store some significant chunk of it. By afford, I mean that the economic cost of off-site storage is less and that the impact on service can also be minimized.

Karl Llewellyn's "threat of the available" [ Karl Llewellyn, "Legal Tradition and Social Science Method—A Realist's Critique," in Jurisprudence:  Realism in Theory and Practice 82 (1962)] is a particular instance of Mooers's Law, which states that an information retrieval system will tend not to be used whenever it is more painful and troublesome for a customer to have information than not to have it. In the context of remote storage, an initial reaction might be to assume that materials stored remotely are more painful to retrieve than materials stored on site and that, as a result, those items stored remotely will be little used. But that is not necessarily true. Ordering a book online is easier than walking to the shelf, as Amazon.com seems to be proving. Only the gratification of immediate desire overcomes it, but that is not a Llewellyn situation.

Remember the precision-recall trade-off. Many Harvard faculty, especially those working in the areas of the core curriculum, used to return from visits elsewhere and comment on how much easier those libraries were to use than was ours. The main reason seemed to be that the shelves of the other libraries were not crowded with previous editions, specialized titles, or stuff in foreign languages. Harvard had too much recall while other collections were more precise.

Many people may not want information; at least they will avoid using a system that gives them too much information. Thus not having and not using information can often lead to less trouble and pain than having and using it.

Actually, it's interesting to apply the Pareto Principle to Penny Hazelton's finding that only 13% of her collection is covered by LEXIS and WESTLAW. If it's the right 13%, the vendors are over half-way home, since it would be uneconomical for them to cover more than the 20% of legal information that meets 80% of our needs.

If that's true, then technology will only replace 20% of our stack space. If you can't get more shelving on-site, don't ignore remote storage.

Finally, remember that we have a fair amount of experience in segmenting our collections by use. We put the most heavily used items on reserve, though even there 20% of the reserve books account for 80% of the use. Many of us have converted large segments of our collection to microform, ignoring the complaints of our users because of the efficiencies. How many of your users would choose on-site microfilm over off-site books? Certainly not all; I suspect not most. In the microform collection, again a few number of titles - you know which - account for the majority of use.

You may not like the 80-20 rule, but it's hard to ignore.

In any event, however, don't forget the most important of all library science rules. Liller's Law of Libraries states that “If it isn't new and doesn't appeal to children, it will be mostly ignored.”