College and Research Libraries


SEYMOUR H. SARGENT 

The Uses and Limitations of True swell 

A recent study at Polk Library of the University of Wisconsin-Oshkosh 
combined statistical and qualitative approaches. Trueswell' s statistiCal 
method was employed, and, after correction of an apparent fallacy in his 
analytical procedure, it was found that the proportion of little-used mate-
rials in the collection was significantly less than shown in similar studies at 
other libraries. An examination of a sample of the little-used books showed 
that , even though Polk Library is a relatively young collection, obsolescence 
is more important than inappropriate selection as an explanation for unused 
materials. 

FOR THOSE INDIVIDUALS associated with 
academic libraries, the current "tax revolt" 
is not a complete novelty. A stock figure of 
the 1970s has been the politician or admin-
istrator who wonders out loud if the library 
is really cost-effective, if acquisitions, par-
ticularly of books and periodicals, couldn't 
be curtailed without any real harm to 
academic programs. Various statistical 
studies-notably the work of Richard 
Trueswell (as popularized by Daniel Gore 
and Stanley Slote) and the Pittsburgh study 
directed by Allen Kent-have encouraged 
speculation that library service as American 
colleges and universities have known it is an 
extravagance. 1 

Hard times usually bring a few benefits, 
of course; and it seems certain that libraries 
will be made wiser if not wealthier by the 
new austerity. If we are to get the funding 
necessary to provide even minimally satis-
factory service, we must be able to justify 
that service with a new thoroughness, using 
every valid procedure that we have to 
analyze and describe every aspect of our 
operations. 

One luxury to which we have been partial 
and that we can no longer afford is a ten-
dency to "choose up sides" on the question of 
whether in managing collections we should 
use quantitative, statistical procedures or 

Seymour H . Sargent is assistant professor, 
Library Science Department, University of 
Wisconsin-Oshkosh. 

416 I 

qualitative procedures that rely on human 
judgment. 

It has clearly been a fundamental assump-
tion that quantitative and qualitative meth-
ods are incompatible. The two most widely 
used texts for library science courses in ma-
terials selection, Carter, Bonk, and Magrill' s 
Building Library Collections and Robert 
Broadus' Selecting Materials for Libraries, 
dismiss the quantitative approach as one 
more dubious version of a "demand" philos-
ophy of collection building. 2 Gore and 
Slote, on the other hand, find the qualita-
tive approach both unscientifically subjec-
tive and a proven failure in practice. 3 

PURPOSES OF A USE STUDY 

In planning a use study at Polk Library at 
the University of Wisconsin-Oshkosh, staff 
decided to try to combine the quantitative 
and qualitative approaches. The study had 
three specific purposes. 

First, we wanted to take another critical 
look at the statistical procedure developed 
by Richard Trueswell and, if possible, use 
that procedure to get an overall understand-
ing of how our collection was being used . 
The Trueswell procedure seemed to be a 
practical way to obtain the kind of informa-
tion we needed; but we were aware that 
many librarians, including some on our own 
staff, were skeptical and even frightened of 
it, largely because of the sensational claims 
by Trueswell and Gore that libraries could 
dispose of more than half their collections 
without noticeably affecting their service. 



However, it seemed to us that the approach 
and the claims could not be ignored. Is the 
procedure sound? Does it lead to the con-
clusions that have been claimed? 

Second, we wanted to follow up a quan-
titative study with a qualitative investigation 
that would try to determine why some ma-
terials were infrequently used. Trueswell, 
Gore, and their followers have generally as-
sumed that to show that a book is little used 
is both to describe a problem and to imply 
the solution: get rid of the book. But it 
seemed to us that it is equally important to 
know whether the book was inappropriate 
in the first place-too technical to be useful 
for our largely undergraduate programs, for 
example-or whether it had become obso-
lete. If we had a large proportion of inap-
propriate books in our collection, we should 
obviously reconsider our selection policies. 
If our little-used books were mostly obso-
lete, on the other hand, our selection policy 
would be exonerated, but the case for get-
ting rid of the books would be more com-
pelling. 

Our third concern was to make a use 
study at an institution of our particular kind. 
Previous use studies have focused on large 
research libraries, special libraries, the li-
braries of small liberal arts colleges, and 
public libraries. 4 • 

The University of Wisconsin-Oshkosh, 
however, is typical of the scores of public 
colleges and universities that evolved out of 
normal schools or were newly established 
during the 1950s and 1960s. Like most such 
schools, Oshkosh is primarily concerned 
with undergraduate instruction. Its library 
has a relativ_ely young collection that has 
been growing rapidly: more than half its 
400,000 volumes have been acquired since 
1969. To what extent can the patterns of use 
in a library of this kind be expected to 
match those that have been found else-
where? More generally, to what extent do 
specific aspects of Trueswell' s procedure 
provide a valid basis for comparing libraries 
with different histories and different pur-
poses? 

FIRST PHASE: 
THE TRUESWELL PROCEDURE 

Common sense tells us that the longer a 
book has sat on the shelf without being 

Uses and Limitations I 417 

used, the less the likelihood that it will be 
used in the future. Richard Trueswell's pro-
cedure, for those not familiar with it, as-
sumes that this principle provides the best 
way of estimating the amount of future use 
that a particular book will get. 

Trueswell was not the first to call atten-
tion to the usefulness of "shelf time" as a 
statistic. Herman Fussier and Julian Simon 
in their use study at the University of 
Chicago had shown that shelf time is a bet-
ter predictor of future use than· the age of a 
book, the length of time it has been owned, 
or its language. 5 It was Trueswell' s particu-
lar contribution to develop a workable pro-
cedure by which shelf time can be related 
to the circulation patterns of individual 
libraries. 

The first step in the Trueswell procedure 
is to record the library's circulation for sev-
eral sample days, recording on a separate 
card for each book checked out the date of 
its most recent previous circulation. · If the 
book has not circulated previously, the date 
that it was first available for circulation is 
recorded, or the best possible approxima-
tion. In our study we recorded student and 
faculty circulation separately for six sample 
days. The days were chose-!1 to be represen-
tative of different kinds of days in our 
academic calendar-regular session week-
days and weekends, out-of-session days, and 
summer school days. In the end, our results 
did not show that the pattern of use varies 
significantly with the kind of calendar day. 
Nor-perhaps more surprising-did the 
pattern of student use seem to differ sig-
nificantly from the pattern of faculty use. 

When we finished collecting the data for 
this part of the study, we arranged the cards 
for each sample day in order of the dates of 
previous circulation. The results were then 
cumulated so that all 1,371 titles wer~ 
classed according to the time since they had 
last circulated, or since they were added to 
the collection. The results are shown in 
table 1. 

The results so far were not very meaning-
ful in themselves, because at ~his point we 
had no way of knowing what proportions of 
the books in the collection had accumulated 
various shelf times. The fact that 64 percent 
of circulation consisted of books that had a 
shelf time of one year or less confirmed our 



418 I College & Research Libraries • September 1979 

TABLE 1 

SIX DAYS ' CIRCU LATION GROU PED BY SHELF TIM E 

Shelf Tim e 
Books Checked 
Out Previously• 

Checked Out 
fo r First Tim e 

Cumul ation for 
Time Pe riod 

C um ulati on as 
Percent of 

T otal Sample (1,371) 

<h30 days 
30-60 days 
00-90 days 
90 days--6 months 
6 months-1 year 
1 year-18 months 
18 months-2 years 
2-3 years 
3-4 years 
4-5 years 
5-6 years 
6-7 years 
7-8 years 
8-9 years 
Before 9th year 

157 
140 
95 

190 
235 
118 
78 
81 
45 
35 
12 
7 
5 
2 
u 

62t 

44 
24 

6 
9 
6 
8 
3 

8 

157 
297 
392 
582 
879 
997 

1,119 
1,224 
1,275 
1,319 
1,337 
1,352 
1,360 
1,362 
1,371 

11 
22 
29 
42 
64 
73 
82 
89 
93 
96 
98 
98 .6 
99.1 
99 .3 

100 
•Jn th e Polk ci rculati on rou tin e at th e time th e d ata were collect ed th e dat e that a book was ret urned fro m ci rcu latio n was stamped in 

th e book. Shelf time here is calcul ated using that dat e. 
t Polk Library r ecord s do not indicat e tl:. e dat e that a book was first mad e avai lable for ci rculat ion ; consequen tly. ligures in this colu m n 

are es timat ed from years of accession and may be off b y 5 pe rcent or so. 
tTh e di stribution of first-tim e and previous checkouts for befor e th e ninth year is probably not accurate as ci rcu lati on records appear to 

be incomple te. 

original supposition that some books circu-
late much more frequently than others. But 
to get a more exact idea of how the collec-
tion was being used , it was necessary to 
look at the circulation records of a represen-
tative sample of the books in the collection . 

A FALLACY IN TRUE SWELL? 

As we were planning the second phase of 
our study, however, we became aware of a 
significant fallacy in Trueswell' s method of 
analyzing his data . In his original work , 
using essentially the procedure we have de-
scribed, Trueswell studied the circulation 
patterns at Deering Library, which is the 
main research library at Northwestern Uni-
versity, and at Northwestern's Technological 
Institute Library. At Deering he found that 
99 percent of the books circulated had ac-
cumulated a shelf time of twenty years or 
less, while the figure for the Tech Library 
was eight years or less. His next step was to 
compute the total number of books that had 
circulated at least once during the previous 
twenty years at Deering and during the 
previous eight years at the Tech Library. 6 

By doing this , he reasoned , he had estab-
lished the size of a "core collection" that 
"should satisfy over 99 percent of circulation 
requirem-ents . " 7 It was this computation 
that led to his claim that "it was found that 

approximately 25 percent of the current 
holdings of the Tech Library should satisfy 
over 99 percent of the current circulation 
requirements. Similarly, a figure of about 40 
percent was obtained for the Deering Li-
brary. "8 This claim , in turn , is the basis for 
Daniel Gore's assertion that "at least half 
the collection may be removed without per-
ceptibly affecting the availability of books 
that people will actually read. " 9 

Trueswell's meaning is not completely 
clear at this point, but his computation is 
either meaningless or fallacious. If he is 
claiming only to have established the size of 
the core, then his point comes down to no 
more than an assertion that, if other factors 
remain the same, about the same number of 
books will circulate during the next twenty 
years as circulated during the past twe nty. 
But if he means that those particular books 
that have circulated during the past twenty 
years are more likely to circulate in the 
future than are the other books in the 
collection-this is the usual interpreta-
tion-then he is wrong. 

He has , in effect , established two 
categories: (1) a core that includes all boqks 
that have circulated during the past twenty 
years (or whatever shelf time has been es-
tablished) and (2) all others . Books. that have 
not circulated at all , it will be noted , are 



excluded from the core regardless of how 
long or short a time they have been avail-
able for circulation. But we have been given 
no reason to suppose that a book that has 
been in a collection for a year or two with-
out circulating is less likely to circulate in 
the future than one that circulated fifteen 
years ago and has not circ~lated since. 10 

To put the point slightly differently, 
Trueswell' s error was to use the shelf time 
of all books that circulated in making his 
first calculation, but then to shift his criteria 
and consider only books with at least one 
previous circulation when defining his 
"core." 

SECOND PHASE: 

SAMPLING THE COLLECTION 

At this point it should be obvious that the 
categories that can be validly established on 
the basis of Trueswell' s procedure are 
(1) the little-used books-those that were in 
the collection previous to some established 
cut-off date and have not circulated since-
and (2) all others. For the second part of 
our study we decided to follow Trueswell 
and assume that those books should be 
counted as central to a library's collection 
that can be expected to account for 99 per-
cent of current use. As table 1 shows, about 
99 percent of the books that circulate at 
Polk have accumulated a shelf time of less 
than 7¥2 years; all books with a shelf-time 
longer than this, consequently, could be 
counted as little used. 

As a sample for the second part of the 
study, we drew 800 titles at random from 
the shelflist and, as a cross check, an 
additional 400 from the title section of our 
divided card catalog. These 1,200 books 
were checked, insofar as they could be 
found, and their dates of most recent circu-
lation were recorded. The results of this 
part of the study are given in table 2. 

In other words, as shown in table 2, only 
about 12 percent of those books we were 
able to locate could be classified as little 
used according to the most reasonable ap-
_plication of Trueswell's procedure. Even 
allowing for our correction of the computa-
tion, this seemed to be a smaller proportion 
of little-used materials than Trueswell had 
found at any of the libraries he had investi-
gated, and so it seemed to imply a distinctly 

Uses and Limitations I 419 

TABLE 2 

RANDOM SAMPLE OF ITEMS FROM 
COLLECfiON GROUPED BY RELATIONSHIP 

TO 7lfz-YEAR CUTOFF 

Percent of 
Number Number 

in Located 
Category Category (n = 1,098) 

Could not be 
located: 102 

Ac~uired since 
lh-year cutoff: 618 56.3 

Purchased before 
cutoff and cir-
culated since: 351 32 

Purchased before 
cutoff, not cir-
culated since: 129 l1.8 

favorable judgment of the relationship be-
tween our collection and the needs of its 
users. 

PROBLEMS IN COMPARING LIBRARIES 

Up to a point, our acquisitions staff prob-
ably can take satisfaction in our results. But 
a consideration of all the information we had 
accumulated up to this point suggests a 
more basic conclusion: the Trueswell proce-
dure must be used with great caution when 
one library is compared with another. 

In his research at Northwestern, as we 
have seen, Trueswell found that at Deering, 
a large research library, the shelf time nec-
essary to account for 99 percent of circula-
tion was twenty years, but at the Tech Li-
brary only eight years. On the basis of both 
these results and common sense, we are 
likely to guess that if the shelf time neces-
sary to account for a given percent of use is 
relatively short, it indicates an uneven pat-
tern of use, with part of the collection, 
probably a rather small part, getting heavy 
use and a large part getting little. If a long 
shelf time is necessary to account for a 
given percent of use, on the other hand, we 
would expect a more even distribution of 
use, with a large proportion of the collection 
being used occasionally. 

But our Oshkosh figures show that such 
an inference would be wrong , or at best 
premature. At Polk we found on the one 
hand that a shelf time of only 7% years 
could account for 99 percent of circulation, 
but on the other hand we found that a rela-
tively small proportion of our collection 



420 I College & Research Libraries • September 1979 

could be classified as little used. The expla-
nation for this initially surprising result is 
obvious from table 2. 

Since the size of our collection has more 
than doubled in the past eight years, the 
collection is relatively young. This rate of 
doubling can hardly be expected to con-
tinue, of course; and as it slows, the propor-
tion of older books in the collection will in-
evitably increase. This, in turn, seems cer-
tain to have two results : the shelf time nec-
essary to account for a given proportion of 
circulation will increase as the average age 
of the items in the collection becomes 
greater, but at the same time the proportion 
of the collection consisting of books that are 
little used because of obsolescence will also 
increase. 

As a readily acquired statistic, the shelf 
time necessary to account for a given pro-
portion of use is an extremely valuable part 
of an attempt to understand and judge the 
effectiveness of library collections. But it 
does not by itself provide an indication of 
what the pattern of use is in a particular li-
brary, nor does it by itself provide a valid 
basis for comparing libraries. 

Not only the age of collections but also 
such factors as the predominant use-
graduate research as opposed to assigned 
readings for undergraduates, for example-
and the amount of use that is made of books 
in various subject areas must be taken into 
account in making comparisons. The dif-
ferent patterns of use that Trueswell found 
at Deering and at the Tech Library, for 
example, presumably resulted at least in 
part from the tendency of scientists and en-
gineers to make heavy use of more recent 
materials. 

WHY ARE SOME BOOKS LITILE USED? 

The third phase of our study was an at-
tempt to determine the most important rea-
sons why some books in the collection were 
not being used. Since they were already 
chosen and made a convenient number, we 
used as a sample the 129 books that had 
been identified as little used in the preced-
ing phase of the study. In carrying out this 
part of the study, we were interested not 
only in the results, but in the practicality of 
using the procedure for weeding. Would it 
require time-consuming research on indi-

vidual titles? Could the classification in-
volved be done with reasonable accuracy by 
professional librarians who did not have 
subject-area expertise? 

We began this part of the study with the 
tentative assumption that we would be able 
to classify the books in the sample according 
to one of three main reasons for their not 
being used: their subject matter might be 
irrelevant to the curriculum of the school 
and to the interests of students and faculty, 
they might be obsolete, or they might be 
highly specialized in their subject matter. 
Early in the study we decided to add a 
fourth major category: out-of-fashion art and 
literature. Later, as we shall see, it seemed 
desirable in practice to elaborate the 
scheme into a few subclasses. 

We found fewer books than expected that 
seemed to fit none of the categories and 
fewer than expected that seemed to belong 
to more than one. In the end, in fact, no 
books seemed to demand classification into 
more than one category. The number of 
"irrelevant" books was also smaller than 
expected, and there seemed to be little 
difficulty in identifying them. Distinguishing 
the "obsolete" from the "specialized" was 
more difficult, and we soon found ourselves 
trying to define "obsolete" in a way that 
would be appropriate to the kind of re-
search library we were studying. 

The Obsolete 

According to one school of thought, of 
course, no book is obsolete from the 
standpoint of a research library, as a use can 
always be found for any given book in some 
hypothetical research situation. But even for 
a research library there is clearly a sense 
in which some books can be counted 
obsolete-a thirty-year-old psychology text,. 
for example, or a book from the 1930s by an 
author of no intrinsic interest that is arguing 
the merits of Prohibition. 

The critical fact about such books, it 
seemed to us; is that they do not have in-
formation about their subjects that is both 
significant and unique: they tell us nothing 
important that is not tOld as well or better 
in other, usually more recent books. In 
some research situations such books can be 
vitally useful, but only as examples of their 
types. And it is characteristic of such re-

,4 



search situations that researchers must have 
access to all books of the type they are in-
terested in, or at least to a large number 
that are known to be representative. 

Historians investigating the treatment in 
American high school history texts of the re-
lationship between the United States and 
Panama, for example, could not be satisfied 
with looking at a few old textbooks that they 
happened to find in a library where they 
were working. At a minimum they would 
need to have access to a substantial propor-
tion of all the high school history texts pub-
lished in the United States during the past 
several decades. For the convenience of re-
searchers as well as librarians and library 
browsers, it makes sense that books likely to 
be used only in this way should be kept 
only in a few large depositories. 

The books that we finally judged "obso-
lete" broke down into two clear-cut sub-
categories. 

The first included "superseded" works-
old textbooks, reference manuals and the 
like, and older editions and translations of 
books of all kinds that had been superseded 
by newer versions. 

In the second subcategory were books of 
"outdated controversy" -books from the 
1930s and 1940s on tariff policies and inter-
national cooperation that had no significance 
as firsthand testimony to events, for example. 

One factor that helped in this part of the 
classification is the direct relationship that 
generally pertains between the extent to 
which a book provides general coverage of a 
broad subject area and the speed with 
which it becomes obsolete. Typically, the 
basic purpose of a general book is either to 
present in a series of broad statements the 
structure of knowledge in a field as it is un-
derstood at the time of writing, or else to 
present in the form of a handbook the most 
commonly used quantitative data in a field. 

In any field where active research is 
going on, some of the broad statements and 
bits of data are continually being proved 
wrong or, more frequently, misdirected in 
emphasis or incomplete. Because they are 
open to correction at so many points and 
also, of course, because they are so much 
used and in such great demand, general 
surveys are usually superseded in a rela-
tively short time. It was relatively easy for 

Uses and Limitations I 421 

us to single out a large, well-defined group 
of general books that were clearly not being 
used because they were obsolete. 

The Specialized 

The "specialized" books· seemed to fall 
into three subclasses. 

First was a group rather like the "out-
dated controversy" subclass in that most of 
its members had once had a broad appeal 
that they had since lost. But they differed in 
that they seemed to have retained sig-
nificant information value for some kinds of 
research, sometimes because they contained 
significant eyewitness testimony about sig-
nificant events, sometimes because they 
were written by persons who themselves 
were of historical interest. We labeled this 
group with the rather inadequate phrase 
"historical sources, etc."; among the items 
in it were The New Reformation, by the 
physicist Michael Pupin; East of Siam: 
Ramblings in the Five Divisions of French 
Indo-China, by Harry A. Franck; and 
Henry Seidel Canby's Alma Mater: The 
Gothic Age of the American C allege. 

Most of the remaining books in the "spe-
cialized" category were technical works of 
scholarship; but it seemed worthwhile to 

· make a separate subclass for the small 
number of books whose failure to circulate 
appeared to be attributable partly to their 
being in a foreign language. 

The results of the third part of our study 
are shown in table 3. 

TABLE 3 

LITILE-USED BOOKS AND 
PRESUMED REASONS FOR LACK OF USE 

Reason 

No obvious reason 
Irrelevant 

· Out-of-fashion art and 
literature 

Obsolete 
Superseded 
Outdated controversy 

Specialized 
Historical sources, etc. 
Technical, scholarly 
Foreign language 

Number 

2 
5 

23 

27 
8 

27 
33 

4 

Inevitably, the classification involved some 
arbitrary decisions. Other persons working 



422 I College & Research Libraries • September 1979 

with the same group of books might have 
developed a somewhat different classifica-
tion scheme. Given the same classification 
scheme, they would probably have come up 
with somewhat different results-but not, 
we think, with dramatically different results. 

CONCLUSIONS 

What does this part of our study show? It 
would seem that the lack of use of the books 
in all our categories except "irrelevant," 
"technical, scholarly," and "foreign lan-
guage" results from processes of change-
the growth of knowledge and shifts in taste 
and interest-that are beyond the control of 
the library. Only items in these three 
categories, in other words, can be taken as 
reflecting on past acquisition policies. 

No acquisition policy, obviously, can 
anticipate future needs perfectly: we can be 
sure of buying nothing superfluous only if 
we refrain from buying much that is 
needed. That only forty-three items-less 
·than 5 percent of our sample of 1,098-have 
turned out to be bad investments suggests 
that, even on the basis of a rather narrow 
criterion of usefulness, our past performance 
has been acceptable. 

Ours is only one library, but we have no 
reason to suppose that our faculty and staff 
have any unusual prescience in acquisitions 
matters. At least for our library and prob-
ably for others of our type, these results 
seem to call in question the recent series of 
claims that a very large proportion of most 
library collections are little used and that a 

large proportion of acquisitions are super-
fluous. 

As we have suggested, as our collection 
becomes older, the proportion of little-used 
materials will almost certainly increase. But 
that increase will result in the main from 
materials' becoming obsolete, and the mate-
rials will become obsolete, of course, 
whether or not new ones are purchased. 

By combining the quantitative and qual-
itative approaches to collection study, we 
have learned things that we could not have 
learned from either taken alone. In a more 
qualified way, our experience suggests that 
such a combined approach is practical for 
weeding projects that must deal with entire 
collections rather than samples. Such use 
would rest on the assumption that the kind 
of books we have called "obsolete" should 
be stored or discarded before the "spe-
cialized." The relatively small subclass of 
"outdated controversy" was difficult to sepa-
rate from the more useful "specialized" 
books, and in most cases we think the ben-
efits would not repay the effort. The much 
larger subclass of "superseded," on the 
other hand, was relatively easy to pick out. 

Rather than adopting a strictly quantita-
tive approach and discarding or storing all 
books that have not circulated in twelve 
years, for example, a library-particularly if 

. it has an automated circulation system-
might be well advised to identify all books 
that have not circulated in six years and 
select for storage, or discard, only those that 
obviously fit the "superseded" category. 

REFERENCES 

1. Richard W . Trueswell has given his research 
with respect to book collections, which was 
originally published in a series of articles dur-
ing the 1960s, an apparently definitive sum-
mary in "Growing Libraries: Who Needs 
Them? A Statistical Basis for the No-Growth 
Collection," in Daniel Gore, ed., Farewell to 
Alexandria (Westport, Conn.: Greenwood, 
1976), p.72-104. Gore elaborated on the ap-
plication of True swell's approach in "The 
View from the Tower of Babel," Library 
Journal, 100:1599-1604 (1975), and in the' 
title essay of the aforementioned Farewell to 
Alexandria, p.164-80. Stanley S. Slote sum-
marizes True swell's research along with his 
own in Weeding Library Collections (Little-
ton, Colo .: Libraries Unlimited, 1975). 

p.67-70, 76--93 . Thomas J. Galvin and Allen 
Kent provided a preliminary report in "Use 
of a University Library Collection: A Progress 
Report on the Pittsburgh Study," Library 
Journal 102:2317-20 (1977), with the final re-
port published in the volume Use of Library 
Materials: The University of Pittsburgh Study 
(New York: Marcel Dekker, 1979). 

2. Mary Duncan Carter, Wallace John Bonk, 
and Rose Mary Magrill, Building Library 
Collections , 4th ed. (Metuchen, N.J. : Scare-
crow, 1974), p.164~5; Robert N. Broadus, 
Selecting Materials for Libraries (New York: 
Wilson, 1973), p.21. 

3. Gore, "The View from the Tower of Babel," 
p.1600, 1604; Slote, Weeding Library Collec-
tions, p.20--21, 35-38. 



4. Trueswell studied two large research librar-
ies , a special library, and the library of a 
small liberal arts college . Herman H . Fussier 
and Julian L. Simon did a more elaborate 
study of a large research library, Patterns in 
the Use of Books in Large Research Libraries 
(Chicago: Univ. of Chicago Pr., 1969). Slote 
described his studies of several public librar-
ies in Weeding Library Collections , p . 76-93 , 
117-65. 

5. Fussier and Simon, Patterns in the Use of 
Books , p.31. 

Uses and Limitations I 423 

6. Trueswell, "Growing Libraries," p .84-85. 
7. Ibid. , p.85. 
8. Ibid . 
9. Gore, ed., Farewell to Alexandria, p.173 . 

10. Fussier and Simon found that, of books that 
have accumulated equal shelf time, those that 
have never circulated have somewhat less 
chance than those that have circulated of 
being used during a given period in the fu-
ture (Patterns in the Use of Books , p .32-33); 
but the difference is clearly not great enough 
to justify Trueswell' s procedure.