ACRL News Issue (B) of College & Research Libraries


October 1996/581

W h e re  ha ve  all the “ lost” 
books gone?

By Edward V. Van Gemert

Projecting overall rate o f loss using 
a sample inventory

I n 1994– 95 the University o f Wisconsin- Madison conducted an inventory of its larg­
est library. The goal o f the inventory was to 
determine a baseline loss rate and to respond 
to concerns expressed by library users— espe­
cially members of faculty library committees—  
that the loss o f materials was inhibiting schol­
arly research. W e sought to replace unsub­
stantiated perceptions with quantitative data to 
guide decisions about collection security.

The Madison campus is the largest in the 
University of Wisconsin system, home to 117 
academic departments, 40,305 students, and 
2,344 faculty. The General Library System (GLS) 
consists of Memorial Library (the central library 
for the humanities, social sciences, and area 
studies), College Library (the undergraduate li­
brary), Steenbock Library (the resource library 
for agricultural and life sciences), and a dozen 
branch or “member” libraries in a range of dis­
ciplines.

T h e  c h a r g e  o f  th e  L o st B o o k  W o r k in g  
G r o u p
The Lost Book Working Group was charged to 
report on the data currently available for lost 
and missing items across the General Library 
System; analyze both objective and subjective 
data to determine the extent o f the problem; 
and recommend actions to the director to alle­
viate the current problems and keep our col­
lections intact.

The membership o f the Working Group in­
cluded staff from central technical services, pub­

lic services, collection development, and the 
member libraries.

M e m b e r lib r a r ie s
Eleven member libraries responded to a request 
for data regarding the number of known lost 
monographs as of 1992– 93. Six were able to 
report figures:

Art 2,154
Biology 935
Chemistry 82
Geology 200 (approx.)
Physics 1,000 (approx. since 1985)
Engineering 1,190 (approx.)

These figures reflect a narrow range of 0.5 
percent to 0.58 percent o f each library’s mono­
graph collection.

Three o f these libraries do an inventory ev­
ery two to three years; one does an annual in­
ventory; one had an inventory in process; and 
one had done a partial inventory in 1991.

M e m o r ia l L i b r a r y ’ s s a m p le  in v e n t o r y
Memorial Library is the largest library on the 
Madison campus with approximately 3.2 mil­
lion volumes in its collections. Since no inven­
tory had been done o f the Memorial Library 
stacks within the collective memory o f present 
staff, the Lost Book Working Group decided to 
simulate an inventory, based on a sample of 
the stack monograph collection to establish 
baseline information on unlocated books.

A search of the literature did not uncover 
any models for conducting a sample inventory 
o f a large academic research collection. Advice 
regarding the methodology and focus o f a 
sample inventory was sought from professor

Edward V. Van Gemert is head o f user services at the University o f Wisconsin-Madison’s Memorial Library; 
e-mail: vangem@doit.wisc.edu

mailto:vangem@doit.wisc.edu


582/C&RL News

The second phase of searching by 
perm anent staff located more 
than half of the volum es that 
student searchers could not find, 
suggesting a serious problem w ith 
inaccurate shelving.

Jun Shao, a sampling expert in the UW-Madi­
son Statistics Department. Based on his input, 
the Working Group decided that a one percent 
sample o f the stack collection (every 100th 
card in the shelf list) would provide an ac­
curate indication o f the overall scope o f 
unlocated books.

With approximately $2,000 in student fund­
ing approved by the director, the first phase of 
this sample inventory of monographs classed 
in the Library o f Congress classification system 
was completed over the summer of 1994. Peri­
odicals and monographs classed in Cutter were 
not included. Cutter classification was used 
before 1953, and approximately 300,000 vol­
umes remain in the Cutter classification.

The one percent sample inventory searched 
for 22,663 titles representing 2,266,300 volumes. 
In the first phase, student assistants searched 
the NOTIS circulation records. Those titles de­
termined not to be in circulation were then 
searched further by students. In this procedure 
students first checked the NOTIS circulation 
database to determine the existence of copy 
holdings and an item record. The item record 
indicated if the volume was charged out, lost, 
or formally withdrawn.1

Since retrospective conversion is not yet 
complete, titles which did not appear on NOTIS 
were searched in the shelflist to determine if 
they were withdrawn or discarded. The titles 
that were not charged out or withdrawn were 
then searched on the shelf in the stacks and in 
the adjacent shelving rooms. Students indicated 
whether the items had been located.

At the conclusion of phase one, 2,142 titles 
(or 9.4 percent) of the sample were not located 
on the shelf. Eight titles were charged to the 
“lost” category on NOTIS, and 42 shelflist cards 
were already marked as withdrawn.

The second phase, a more methodical, com­
prehensive check by permanent staff in user 
services, began in December 1994 and was com­
pleted in March 1995. Permanent staff are ex­

pert in searching the possible 
locations within the library. They 
took into account misshelving 
and searched the adjacent stack 
areas. They thoroughly under­
stood oversized items which are 
shelved separately, and were 
knowledgeable about various 
reading room locations through­
out the library.

In this phase, permanent staff 
found many of the books not located by stu­
dent assistants. The original list o f 2,142 miss­
ing titles was reduced to 973. This represents 
4.3 percent unaccounted for in the sample of 
22,663. We consider this to be our net figure 
for unlocated materials.

H ig h e s t  a r e a s  o f u n a c c o u n te d  fo r  
titles
A difference o f means test was applied to the 
data to determine if certain areas had statisti­
cally significant higher rates of loss.

This analysis determined a rank order from 
high loss areas to low loss areas within the col­
lection. A complete breakdown can be obtained 
from the author upon request.

As noted, the overall lost and or missing 
rate was 4.3 percent. The top four high loss 
areas were:

HD 1-4730 Economic History
and Conditions 10.9%

KF Federal Law 10.3%
PL 1-3316 East Asian Languages 

& Literature 9.8%
L Education 9.7%

These subjects experienced more than twice 
the average rate of loss.

Using the results from the one percent 
sample inventory; we projected the following 
figures for the LC collection as a whole:

• 97,300 items would not be on the shelf 
or otherwise accounted for;

• 4,200 items would be formally withdrawn 
from the collection based on shelflist informa­
tion; and

• 800 items would be charged to lost on 
NOTIS.

T itles m is s h e lv e d
As noted above, the second phase o f search­
ing by permanent staff located more than 
half o f the volumes that student searchers



October 1996/583

could not find, suggesting a serious prob­
lem with inaccurate shelving. Therefore, 
Memorial Library’s permanent staff shelvers 
w ere asked to report items found to be 
misshelved in the course o f their regular 
work. These items were then checked against 
the circulation system to determine if search 
requests had been placed on these titles by 
patrons. In a two-month period, 1,225 items 
were reported as misshelved. This indicates that 
a large number o f items are misshelved, and 
although not entirely relevant to the sample 
inventory, data were used by the Working 
Group to form a recommendation.

C o m p le te  in v e n t o r y  o f  a n  LC  s u b c la s s
In response to concerns expressed by library 
users and librarians regarding unaccounted for 
tides in the Spanish literature portion o f the PQ 
classification, a complete title-by-title inventory 
was undertaken. The first phase studied 19th- 
and 20th-century Mexican literature (PQ7297- 
PQ7298.36).

This classification was recommended by fac­
ulty and selectors alike. There are 4,600 titles 
in this subclass. The findings are shown be­
low:

169 (3.7%) charged out to patrons
3 Charged to lost
2 Charged to bindery and 

preservation collection
27 Withdrawn per the shelflist

159 (3.4%) not located

The Lost Book Working Group concluded 
from this title-by-title inventory data that sub­
class PQ7297-PQ7298.36 was not, after all, a 
high loss area. In fact, it turned out to be lower 
than the average loss rate, contradicting user 
perceptions.

S u m m a r y  c o n c lu s io n s  a n d  
r e c o m m e n d a t io n s  to th e  d ire c to r
The Lost Book Working Group recommended 
that the following steps be taken:

1) Another sample inventory should be done 
to determine an annual loss rate.

2) Shelf reading should be done at least an­
nually in the LC classes.

3) Special collection budget allocations 
should be made to selectors who can docu­
ment losses in their subject areas.

4) An automated inventory process should 
be developed.

5) The replacement fee for unreturned and 
lost materials should be increased.

6) Libraries should purchase up-to-date se­
curity detection systems.

C o n c lu s io n
The Memorial Library inventory demonstrated 
the feasibility of using a one percent sample to 
gather data on loss of materials.

The costs in student wages and permanent 
staff time proved to be affordable and reason­
able. The 1994—95 sample inventory provides 
baseline data only. Since materials may have 
disappeared over a period o f decades, we can­
not calculate a current loss rate until we under­
take further inventories, as planned for 1997 
and periodically thereafter.

A one percent sample provides sufficient 
data for decision-making on a broad scale; it 
does not, of course, identify all missing items 
at the title level. Where there is reason to sus­
pect systematic losses in a subject area, a full 
inventory can be conducted to identify titles 
for replacement.

Au. note: I’d like to thank the students and 
permanent staff o f Memorial Library’s User Ser­
vices Department for their contribution to the 
completion o f the inventory; members o f the 
Lost Book Working Group; professor Jun Shao 
for his consultation on statistical methods and 
sampling; Gregg Gunderson for his consulta­
tion and data analysis; Susan Searing, the asso­
ciate director for public services; David Henige, 
the African studies bibliographer; and Don 
Johnson, the editor for the General Library Sys­
tem, for their careful reading and editing of 
this article.

N o te
1. “Missing” is a term used by circulation to 

indicate that an item cannot be located in the 
stacks. After searching, the item may or may 
not be located.

“Lost” is a term used when an item is 
searched for and not located after six months.

“Withdrawn” is a term to indicate that the 
item has been declared lost. Circulation noti­
fies central technical services to update the 
record for unavailability o f the item. It may even­
tually be located or returned.

“Discarded” indicates that the library has 
intentionally removed the item from the col­
lection. For example, the item is removed for 
preservation reasons. ■