to bar code or not to bar code

[Felix Sperling]

Dear Taxacomers:

        The following long rumination is a request for advice on bar code
labels for insects. I hope it also generates some discussion.

        We are pondering the desirability of using bar coded labels to
identify individual specimens in a database for the Essig Museum of
Entomology. We have 4.5 million specimens and no realistic prospect of
back-labelling them all, but we would like to start labelling incoming
material and taxonomic groups that are part of ongoing revisionary work. We
are not debating the merits of giving individual specimens their own
identifier and entering the information into a growing database (we are
convinced of its value) but rather how best to affix the identifier. More
specifically, do we use bar code or not?

        I have asked around about some of the bar code systems that are in
place (or not) (thank you to Felipe Onoro, Norm Johnson, Steve Ashe, Rob
Colwell, Gail Kampmeier, and Mike Sharkey). It looks like the system we
would go with would be a label that is .315 x .56 inches (8 x 14 mm), using
code 49 and two or three stacks of code, depending on how many digits are
needed, with the number printed below the bar code. These can be made up by
Intermec Pre-print (Fairfield, Ohio, ph. 1-800-221-8314) on polyester paper
for $25. per 1000. plus $700. setup for each order less than 250,000. The
reader would most reasonably be an Intermec 1545, and all hardware costs
including cables but not computers should be less than $2000. There is no
way that we can put in a really big order of labels, but we could scrape up
the resources to get hardware and a supply of labels for the next year or
two for $4000.

        The question is - is it worth it compared to the cost of just
printing up sequential numbers on good fiber paper with a laser printer and
typing them back in whenever the specimen needs to be reprocessed?

        Here is my assessment of the pros and cons:

Merits

1. Bar coding can save time. Although bar coding is approximately as time
consuming to set up as an unadorned numbering system would be, real time
savings can acrue when specimens are processed on subsequent occasions.
Collections with industrial volumes of loans transactions are probably the
most important beneficiaries of this type of application (e.g. INBio),
although the recipients of loans are also sure to be grateful for anything
that actually speeds up getting a loan. In addition, there are likely to be
fewer errors than if numbers need to be typed into the loans database.

2. Bar coding is "high tech". Because bar coding is seen as high tech by
deans and their ilk it is deemed good and is therefore funded. High tech
gives the appearance of quality and progress to anyone disinclined or
unable to actually judge it on its scientific or practical merits. In a
more charitable vein, programs that make an effort to keep up with new
technology even when its benefits are neutral are the programs that are
most likely to make substantial strides when technology that is useful
comes along. Such usefulness is very hard to judge ahead of time. In any
case, there is plenty of evidence that technology attracts money regardless
of merit.

Demerits

1. Bar coding may not save any time. For collections that process smaller
amounts of loans, say in the order of 50-100 loans per year, averaging
100-200 specimens per loan, it is not obvious that there is much time
saving. Typing numbers into a database and then checking them again,
especially if you can do it without moving the specimen, is not necessarily
less efficient than using a bar code reader at that scale. I am not aware
of any bar code systems where you don't have to a) take specimens out to
run downward-facing labels over the bar code reader, or b) need to
reposition specimens that have upward-facing labels widely enough so that
you can the point the reader at bar code labels without confusing the
reader when it hits neighboring labels. Even if you save 30 minutes for
each of 100 loans with a bar code system, then the time saving of 50 hours
has to be compared against the cost of paying a student for that time (max.
$12./hour = $600. per year) and the yearly cost of maintaining a bar code
system.

2. Bar coding may create hidden continuing costs. It takes time to set up a
bar code system in the first place, although that should be a one-time
expenditure. Moreover, there is likely to be yearly turnover in the student
workers who use the bar code reader, and I am told that it is no trivial
matter to train a new person to use a reader effectively. This training
time has to be compared to the simpler task of telling someone to type and
recheck numbers in a loans database. More worrisome is the possibility that
bar coding can create costs that continue to eat away at operating budgets.
Because an entomology collection needs tiny labels that are at the low end
of resolution of readers, I have been advised that it is foolhardy to try
making labels with a laser printer rather than having them made
commercially. Such labels cost $25. per 1000 at Intermec, which seems
reasonable enough until you realize that you also need to pay $700. (up
from $500.) for setting up the printer for each order, even if it is just a
reorder of a previously ordered series of labels. Somewhere along the line
you need to consider that even bar code readers can break down or wear out,
and that will also cost money.

3. Bar coding could create a white elephant. Most of the bar code labels
that are affixed to specimens will never be read again in the next ten
years, but are very likely to be of interest within the next 100 years.
This is just one of the facts of maintaining a long term storage depot for
odd things that might or might not be valuable some day. On the other hand,
I find it hard to believe that something significantly better than bar
codes won't come along in the next ten years (e.g. microchips or direct
text readers). Any sensible bar code label would of course have the number
printed on it along with the bar code, and therefore it would not be
useless, but in order to continue to gain the benefits from having bar code
as part of the label you would probably have to continue to maintain your
old bar code reader system. Alternatively, you might want to just abandon
it and cut your losses, or you may be lucky and be able move to a direct
text reader that can still read bar code as well. My bet, based on
technology turnover patterns in general, is that you won't get lucky
without having to make another significant investment.


Well, these are the arguments as far as I can tell from pondering the
problem over the last couple of months. I would be very grateful now to
hear from anyone else who has experience in the use of bar code for insect
specimens.


 *  *  *  *  *  *  *  *  *  *  *  *  *  *  *  *  *  *  *  *  *
 Felix Sperling                   office: 510 642 4296
 Assistant Professor              lab:    510 642 7410
 201 Wellman Hall                 fax:    510 642 7428
 ESPM - Insect Biology
 University of California         sperling at nature.berkeley.edu
 Berkeley,  CA  94720-3112