Changes in the use of literature with time - obsolescence revisited

INTRODUCTION

In 1974, an article by Sandison and myself was published with the title "|Obsolescence' and Changes in the Use of Literature with Time" (Line & Sandison, 1974). It reviewed all the literature on the subject that we could find and came to some interesting and unorthodox conclusions. It has been widely cited and has become something of a classic. This article looks selectively at subsequently published literature that deals with the issues we discussed earlier with particular reference to the practical implications for libraries. It goes on to examine the relevance of the concept of obsolescence in a possible future world where libraries are "virtual" rather than "actual." It is useful to start by quoting extensively from our conclusions:

The study of obsolescence has been bedeviled by superficial approaches

to a highly complex situation of interlocking factors. It is most important,

first, to be quite clear whether changes in library use, in reference/citation,

or in the value and interest of knowledge are being considered, and

to ensure that conclusions about one of these are not drawn from data

on another without adequate evidence for their interrelation. Secondly,

great care must be taken to seek out all age-related variables that can

bias the data examined, and to make suitable corrections for them all.

Thirdly, full allowance must be made for the extremely high variances

in all the variables concerned, and to ensure that conclusions are not

being drawn from differences which lie within the experimental errors.

With these provisos kept clearly in mind, studies of the relation between

use and age are of considerable potential interest, even if much of this

interest is theoretical.

.... Most estimates of "obsolescence rates" given in the literature derive

from synchronous use-frequencies in which the growth of the literature

exaggerates the apparent age relation. It is not yet known ... what factors

determine which items in a collection will be read or cited, and the

relative importance of age among them all.

What in practice library or information systems require is a method

of estimating the relative probabilities that particular items will not

be used. There appear to be such enormous variations between and within

subjects, titles, and age groups, that only a probabilistic approach is

likely to lead to useful results....

On the evidence so far available, age seems to be a rather poor criterion

for discarding or relegating to less accessible storage .... It may be sensible

to discard either after about three years (so avoiding binding), or not

at all. It also seems likely that discarding whole runs of less used serials,

whether currently received or not, is more cost-effective than discarding

older volumes of all serials taken. A library may therefore do best to

see (a) which journals are "dead" (whether because they have ceased

publication or because the library has ceased subscribing), and to consider

whether they can be discarded, (b) which journals receive little use of

their current issues, and to consider whether they can be cancelled and

discarded, and (c) whether there are some journals that, although

currently used, fall off so completely in use after three years that they

need not be bound or retained after that time....

It appears then that "obsolescence" is not a concept of which librarians

can make much practical use....Any librarian who attempted to weed

on the basis of assumed "obsolescence", or of other studies which

purported to show it, would be very unwise.

It is not known how useful references and citations may be as indicators

of use probability, nor how usage patterns differ between libraries,

whether of the same or of different types.

Not only are precise measures needed of changes in use with age,

but relevant measures. The most precise measures for each age group

would be, for library uses, uses per available item per user, and for

references, references per citable item per potential reference. However,

these measures may be of less practical use in the decision whether to

retain or retire....than relative uses per metre....(pp. 318-20) Before the conclusions, we advanced several hypotheses:

Literature may decline in use faster when

(a) it deals with data of ephemeral relevance....

(b) it is in the form of a "report", thesis, "advance communication"

or preprint....

(c) it is in a rapidly advancing technology.

Literature may decline in use more slowly when

(a) it is descriptive (e.g., taxonomic botany...

(b) it deals with concepts (e.g., philosophy...

(c) it is critical (e.g., literary criticism... ). (pp. 317-18)

We might have added that in the humanities and social sciences, the use of "raw" material--historical sources, literary texts, and so on--declines in use (if indeed it declines at all) more slowly than that of "secondary" literature which interprets it. Most histories written more than forty or fifty years ago are of no interest except to the historiographer, but the original sources are still used by today's historians. Similarly, most older theological works are dead to all intents and purposes, while Migne's Patrologia Graeca and Latina have recently been issued on CD-ROM.