Stone Age thinking at the speed of light
AACR2, MARC and other
dinosaurs
JOHN JOLLIFFE
Transcribed
from a cassette tape recording1 and annotated by Geoffrey Neate and
David Helliwell. Thanks to Terry Belanger2, Kay Guiles, and Robin Alston
for valuable information on the persons and events mentioned in the lecture.
March 2007.
I welcomed the opportunity of preparing this
talk before I welcome, time permitting, the opportunity of attempting to
crystallise my ideas on some topic which has been growing in importance in my
mind over a period. I’m not the sort of performer who dons cap and bells and is
happy to give for the fiftieth time a recitation of the Seven Ages of
Librarianship. After suggesting this talk to Terry3, and then seeing
the other speakers here included in his programme, I realise that I might have
perhaps better chosen a different topic, another one on which I must soon
crystallise some thoughts, and that is on certain aspects of book pricing and
distribution on the Continent during the sixteenth century, but I’m afraid
you'll have to wait for that one.
The
conversation I had with Terry last fall, which led to the idea of giving the
talk, had been on matters nearer the present idea, and so I naturally thought
of this as being the one that he would like me to dilate upon. I must apologise
in advance for some comments which may be not generally understood because they
involve echnicalities of computer processing or of programming. Such comments
will, however, tend to reinforce a proposition which has been commonly accepted
since the first days of using computers in libraries, namely that librarians
should know more about computers, not simply from the outside, knowing the uses
to which they are put, but from the inside, knowing by using them at practical
level what they are capable of being used for. A very long time ago, I was
introduced to a statement by Ada, the Lady Lovelace, daughter of the poet
Byron, mathematician, and supporter of Charles Babbage, the inventor of modern
computer programming.4
Referring
to the machine which Babbage was designing and building with government support
over about twenty years, and without success, she said, “It will do whatever we
know how to do.”5 Now the meat in this proposition lies in the words
“whatever we know how”, and I’ve found it to be true in nearly twenty years of
work with computers. In this time I have also met remarkably few librarians who
knew either the statement, or its truth. I have however met many who have
worked with computerised library systems who are unaware of this truth and of
the obligation it lays on us, not on the machines, for the limitations of the
systems which we design and implement.
Looking
back to the early 1960s, it is possible to discern a simultaneous development
of two features, which especially when they have interacted, have begun to
change the practice of libraries. On the one hand there has been the continuous
development of the most sophisticated and flexible instrument ever designed by
the mind of man – the computer – and on the other there has been the
introduction into library practice of ever more restricting codes, and ever
less flexible standards. This is ironic.
A further irony is that in recent years it
has sometimes been said in justification of a new restriction that this is
because of the requirements of the computer. I will put forward a thesis, which
you may be relieved to hear I will not attempt to justify word by word, that
the thinking behind the use of computers in libraries is out of date, and
possibly has always been so, and that the opportunities for newer modes of
thought and newer methods offered by technological advances have largely been
ignored. It would be possible to argue, as some computer specialists are now
arguing, that by 1965, the computer structures based on the Von Neumann logic
of the 1940s, and on the limitations imposed by the earliest component of the
computers, were out of date. I shall return to this briefly later on.
For the moment I want to remain with library
system designers, rather than computer designers. My first contact with
computers was in 1964. At that time I was working in the library of the British
Museum6, and the head of the library, Robert Wilson, had been asked
by Sir Frank Francis, who was then the Director of the Museum, “What are you
doing about this new idea of using computers in libraries?” So I was given
fourth refusal of the question, “Would you like to find out about this idea of
using computers in libraries?” I said “yes”, and the story should go “I have
never looked back.”
In
the following year, together with a colleague from the Library, I came to the
United States for the first time and for a period of three weeks the two of us
were almost in daily contact with those in the Library of Congress who were
planning the MARC project. I do have a longish memory, although I hope I’m not
yet a fossil. In those days Snyder was not Henry [laughter] and ESTC North
America, but Sam and the LC Systems Office.7 He was Mrs Avram’s8
colleague and superior. What has been fossilised from that time is MARC itself.
Then they were planning an experimental distribution of cataloguing data on
magnetic tape to, I seem to remember, only fourteen libraries, and for a
limited period of time, perhaps six months.
The
inevitable happened. Those first thoughts, subject in principle to revision and
improvement in the light of experience, are still with us. The libraries
participating in the experiment, which had invested so much effort in
programming, and which had altered their systems to depend on the continued
availability of cataloguing in the experimental form, were naturally reluctant
to encourage any radical modification of the MARC structure. In consequence,
the experimental nature of the MARC 1 project really did not come to any
conclusion. Certain features, which even in their design phase in 1965 were
acknowledged to be weaknesses,
necessary
features have yet been added. are still with us. Effort was concentrated on
modern books, indeed on modern books in English, and all the arguments of
myself and my colleague for a wider specification, within which our problems at
the Museum of books of all dates and in all languages might be accommodated
were listened to, but set aside in favour of the experimental format, and the
coding on to which it was later necessary to graft, and not always easily,
further features to accommodate non-English and non-modern books. Not all the
necessary features have yet to be added.
At
that time, generality was applied to the MARC record itself. It was to be
independent of the medium of transmission. And so it received the
character-string format, even for numerical references within the record, which
is still a feature of the international standard, though to the best of my
knowledge no-one ever wished to transmit MARC records over telegraph lines to
teleprinting equipment.
Another
relic of this early preoccupation with, in effect, validation is the occurrence
of field-terminator characters, as well as machine-generated pointers to
fields, which give the length including the terminator. Thus to get the data
from the field, one has to start by subtracting one from the length.
Belt-and-braces indeed, both supplied by machine!
And
how did this happen? The logic of circumstance and the logic of need only
partly accounted for it. One must remember that in those early days, the only
solid input from the library side was from Mrs Markuson.9 perhaps I
can depart to a small anecdote about Mrs Markuson here, which is not intended
to disparage her.
During that three weeks we were in
Washington, there met an organisation called COLA – and I don’t know if it
still exists. It stood for Committee on Library Automation, and it had
apparently started as being a talking group in the corridors and bars of ALA,
being those people who were already starting to try to use computers in
libraries.10 By the time we met it had become a little more formalised. It had
a chairman, though it did not yet have an agenda and minutes, and we were asked
if we would like to sit in as observers, which we did. And then one evening Mrs
Markuson provided entertainment at home for all the group, including ourselves.
When we arrived, she met us by saying, “Well, after the meal you and Sandy
Cain11 can tell us about your plans for the British Museum.”
We were saved from this in effect by the fact
that what was served as wine with the meal was sherry, and Sandy Cain and I,
finding it was sherry, took it very slowly. Others drank it as if it was
Algerian plonk [guffaws from the audience] and after dinner when we came to
give our talk, I think we were probably the only two in the room who were still
awake.
Now, as I say, in those days there was only
Mrs Markuson who was from the library side, and the dominant people had come
from computing, and as far as I remember, computing of a very strange form, in
code-breaking, here and there. They tended to bring with them some of their
previous experience, but they also saw the library world in concrete terms. And
what they saw as the most obvious manifestation of library information was, if
you’ll pardon the expression, the 5" x 3" card.
Cards
were still in vogue in computer systems. You’ve only to look at films of the
1960s to see that whenever a computer or a retrieval system has to be
emonstrated, what you actually see is a punch-card collating machine from which
someone draws the one card which will indicate whatever it is.
In
those days the terminal as an input device did not exist. Key to magnetic
medium input devices for off-line data capture were in their infancy, and since
they’d been manufactured by the typewriter division of IBM they were naturally
incompatible with the computer division of IBM (which caused a certain amount
of difficulty here and there). The punch-card was almost everywhere dominant.
And so it was natural first to think in fixed field ways, and the leader of the
MARC record is still a punch-card. It was natural also in those days to think
of asking library staff to prepare their records for the machine in block
capitals, on 80-column wide coding sheets. IBM had not yet admitted the
possibility of lower-case letters, and never has adopted the international
character-set standard, which is built round the common alphabet.12
But the library card gave the model for the order of presentation and tagging
of data elements, and accounts for the curious continuing presence of both
fields and sub-fields in the record, which actually requires two modes of
access within the same record to certain data elements. You use a pointer to
find the field, and then you have to do a character-by-character search through
the field to find a sub-field. Why not do one or the other? Why do both?
The
MARC record contains these subfields because on certain lines on the card you
will have, for example, an imprint, and there it all is – place, publisher,
date, and size, so the line is a field, but nevertheless we want to get at
these individual things, so we chop it up within the line. The notion of having
several fields on the same line was impossible of conception.
The
MARC record is not totally unchanged since the beginning. The separation of the
data from the pointers, so that the pointers refer to the area of the data
only, and not to an area relative to the start of the record, is a modification
which was intended to permit changes of the record by addition or deletion of
data without the necessity of changing all the addresses in all the pointers.
But the pointers still appear in numerically ascending order of tag, but there
is no necessity for this. And in spite of the generality of the pointer system,
the data fields are also concatenated
in
tag order. Even where it is possible, no two pointers point to the same data.
On the typed card, if “London” is needed twice it is typed twice because it is
in two different places. In the MARC record, it would be perfectly simple to
have two pointers pointing to the one word “London”.
For
some data elements there is still no provision, for example, bibliographical
format, bibliographical collation, a reference to the bibliography in lieu of a
standard book number, for example a reference to STC.13 But here I
start to tread on different ground. It might be unreasonable to expect MARC to
provide for elements absent from AACR, just as it may be unreasonable to expect
MARC not to include elements in AACR for which there is no logical
justification when applied to certain kinds of book – height of copy, for
example.But before coming to AACR, a brief diversion into the ISBD and the
ISBN.
I
remember the time when the ISBD was no more than a gleam in Michael Gorman’s14
eye. I remember why it was proposed. The notion was that with the growth of
MARC-based national bibliographic centres, it would be advantageous if
bibliographic records from countries which had a national bibliography but did
not yet produce MARC records on magnetic tape, if their records could be put in
such a form that they could be typed into the computer in the more advanced
countries, and put through a single program called an automatic format
recognition program,
which
would turn the records into fully tagged and structured MARC records. The idea
was not absurd. The implementation has been.
Basically,
only those advanced countries which produce their national cataloguing in MARC
form have adopted the ISBD. This is still not quite absurd. What is absurd is
that those countries, which because they had their records in machine-readable
form, could print them out in any way they wanted – backwards, putting “God
save the Queen”, twelve asterisks, or the text of the Gettysburg address if
they so wished as the punctuation between title and imprint – have adopted the
barbarous and illiterate punctuation of the ISBD.
Because they did not realise the real
independence of the input record, the processing record, and the output record,
the intended input record has become the normal output record! Not a great
amount of straight thinking there.
The
ISBD also illustrates another feature of some recent progress towards
standardisation: the tendency for one standard to prejudge an issue not yet
itself standardised. In this case the ISBD depends essentially on the
acceptance of AACR as an international cataloguing standard. Put another way,
this means if you don’t like AACR, then you won’t want the ISBD.
Another
example of this sort of prejudicing of issues, perhaps a little clouded by real
international politics, comes from the potentially unexciting field of
character sets for bibliographic use.
There
is an international standard for the Roman alphabet, and the characters such as
numerals and punctuation commonly used with it. This standard is virtually
identical with the American ASCII code. No-one has ever asked, and no-one has
ever thought another order would be possible; but no-one has ever asked why the
letters in this code are in alphabetical order – they are, it seemed a
reasonable way to do it. In both these codes, the numerical values assigned to
letters are in two blocks, one for upper case letters, and the other, with
higher numbers, for lower case letters. No-one ever thought why you should do
this, in that particular order, but it was done. A random choice, if you like.
Within
each block the normal alphabetical order is followed with each number one
greater than its predecessor. Thus capital “A” has the value 65, capital “B”
66, and so on, while the lower case “a” has 97, lower case “b” 98, and so on.15
This makes sorting of English and most western European languages quite
straightforward. But not all – Spanish, Welsh, Danish, and other small
countries of that sort cause some difficulty.
A considerable amount of work has been done
over the years to provide additional standard sets for other groups of characters,
the Greek alphabet, for instance, a further roman set containing common
accents, and characters such as the “ć” and “oe” digraphs. One proposal came
forward at the International Standards Organisation for an extended Cyrillic
character set.
Now
the Cyrillic alphabet presents a problem. It has more characters than the roman
alphabet, so that it is difficult to squeeze it into a set of numbers of the
same size. In addition there are in use in Bulgarian, Macedonian and
Serbo-Croat some characters which do not appear in the Cyrillic alphabet used
in Russia. There is also the need to provide for a certain number of characters
in the Cyrillic alphabet which were discontinued after the revolution but which
still appear in nineteenth-century Russian book titles.
The extended set was intended to include
these additional characters for Bulgarian, Macedonian and pre-Revolutionary
Cyrillic, and not the characters which the Russians use when writing in
Cyrillic the non-Slavonic languages of Soviet Asia – a further set would be
necessary for that in due course. Now this character set had been prepared by
the Russians, and was based on their own domestic standard, called GOST (GOST
is the name of their standards organisation)16. The International
Standards Organisation accepted it on the nod, and it was circulated to
national standards organisations for voting – ISO has a democratic structure.
Majority voting counts. Indonesia and Thailand together can
nullify
the United Kingdom and the United States. For some reason best known in the
Library of Congress, the United States representatives of ISO acquiesced in
this procedure. Since it was in the euphoric phase just after the Helsinki
agreement17, perhaps no-one wanted to be beastly to the Russians.
The voting was interesting. Of the sixty or
so national organisations who voted, only one negative vote was recorded –
that, I am proud to say, of the British. We voted against it on these grounds.
First, that it claimed to be an extension of a set which had not itself been
circulated as a standard. Second, that the block of numbers assigned to lower
case letters was lower than that assigned to upper case letters, thereby
reversing the procedure in all previous sets. And thirdly, that the
alphabetical order normally assigned to the letters in the Cyrillic alphabet
had not been followed. Now since we had the last two of these objections also
to the parent GOST set, we did not wish to let this slide into acceptance and
debated, because then the parent set would then have had to go through on the
nod because the extended set had already been accepted.
Now the committee of which I am chairman has
been subjected to considerable pressure by the standardisation bureaucracy to
change our vote, for the sake of international agreement, harmony, and
solidarity. Only one of the bureaucrats we have dealt with has given any weight
to our intellectual arguments. We are nevertheless going ahead with the
production of a British standard for Russian Cyrillic, which we shall propose
to ISO as a model.
A last word on this to illustrate part of the
problem.
The first three characters of the Cyrillic
alphabet are, as it were, “A”, “B”, and “V”.18 The “B” doesn’t look
like anything, but “A” and “V” look like “A” and “B”. The GOST standard
assimilates where possible Cyrillic letters to the values given to similar
shapes in the roman alphabet. That is, 65 goes to their “A”, 66 to “V” (because
it looks like “B”), and “B” – their “B” – to a much higher value because at
that stage something like Q, for which there is no equivalent. Thus sorting is
no longer simple. Because the sorting order values – alphabetical order –
differ from the numerical code values. And in objecting, we are trying to
ensure that in England anyway, libraries don’t have to go off into another
sorting procedure when they hit Cyrillic.
Another
standard which has had much more acceptance, certainly in Britain, is the
International Standard Book Number.
Now
although the spelling mistake was not an invention of the early printers, it
seems reasonable to say that they and their successors have institutionalised
it and disseminated it in a way that the mediaeval scribes and their
predecessors could not have hoped to emulate.
Now
the designers of the standard book number did have error and its avoidance in
their minds – why else did they produce the elaborate means of supplying the
check digit at the end? Yet they seem to have had total confidence in printers.
Any
bibliographer could have warned of the practices which were likely to arise in
the use of the standard book number. Indeed, any bibliographer might have
addressed the problem, what is a standard book number. This is not just an
exercise in philosophy. I would expect most librarians to say, that the
standard book number is a number printed in the book which is a unique
identifier for the edition of which that book forms part. Now please don’t
question if I’m right.
Now
the standard book numbering agency in London, and the producers of British
National bibliography as well, hold that the standard book number is a number
assigned to an edition whether printed in it or not, and regardless of any
apparent standard book number (differing from the assigned number) which is actually
printed in the book.
From
this conflict of definitions, and from certain publishing practices, for
instance, perhaps the sensible one of providing a different standard book
number for a hardback and a paperback edition, the less sensible one of sometimes
providing the same standard book number for all four volumes of a four-volume
work, sometimes insisting on providing four different numbers, sometimes
providing a new standard book number for a revised edition and sometimes
sticking to the old number, sometimes even using a standard book number for a
new work because the work to which it was originally assigned is out of print
and therefore the number becomes available – all these things exist, and this
through publishing practice. But of course the printer has the last laugh.
We tried to urge the British National
Bibliography to have an indicator – and there is provision for indicators in
the MARC record – an indicator against the standard book number which will show
whether it was actually present in the book. Or perhaps a range of indicators,
one to show that the standard book number was printed on the back of the
title-page, where one expects to find it, one to say it’s on the dust jacket
only, and a third one to say this is a number that has been assigned by the
agency but doesn’t appear in the book.
But it’s not even quite as simple as that. I
have on my shelves two OUP books in which I can see the standard book number,
but the standard book number in normal form is nowhere printed in it. And what
OUP have done there is take the numeration part of the standard book number,
leaving off the check digit, and leaving off their own prefix, and use that
number in the signatures. Now, is the standard book number in that book or not?
As I say, it’s the sort of question a bibliographer could reasonably address.
We also urged them, as part of descriptive cataloguing, to record as standard
book number what the printer had actually printed, regardless of any inherent
errors. This latter they now do, though this caused them a certain amount of
difficulty.
At
one stage we wanted to test their card service, and we took two MARC tapes a
month apart, and by applying a randomising procedure to it we printed out from
these two MARC tapes two hundred standard book numbers from each, and we
printed them out on the teletype and we simply tore the paper off the teletype,
and with a covering letter saying “Please supply these cards” sent it off to
the British National Bibliography.
In
each case we had a letter back accompanying 198 cards saying “Please check the
other two”. Well, untouched by human hand! At no stage had we transcribed what
was on the MARC tape. It was conceivable there was an error in the teleprinter,
but it didn’t seem very likely, and being reasonably well placed in Bodley with
the bulk of British publications coming into the Library, we shot it off and
looked at the books themselves, and sure enough, what was on the teleprinter,
what indeed
was
on the MARC record, was what was on the book, and it was true the check digit
didn’t check. So although they had forced these things someway, somehow,
through their validation procedure to get the record sitting on the file, it
was not possible similarly to force through the validation procedure any string
which would actually match it, from which I concluded that perhaps one per cent
of the MARC records were never going to be available to anyone.
Now, AACR is currently undergoing revision.
What number will emerge is hard to say – a series whose first two terms are 67
and 2 can have almost any number as its third term. After the fuss about the
cost of change from 67 to 2, one is surprised to find that undaunted, they are
still working towards perfection.
One
of the more honourable scars I bear comes from a rebuke by John Rather19 for
saying that the base of all cataloguing systems is economic. I meant that
cataloguing systems were a compromise between what the library can invest in
terms of labour for creation and maintenance of catalogues, and the utility of
the catalogue to the user of the library. He objected to the notion that
cataloguing rules were anything other than some branch of philosophy. Or was it
theology? Certainly his party were in control, even though they shifted their
ground over the years.
AACR2
abandoned one of the tenets of the Paris conference on cataloguing principles,20
one of the few tenets which I thought was AACR2. The report was originally
published by IFLA in 1963, and was photolithographically reproduced with an
additional introduction in 1981. It was actually an advance on Panizzi, namely
that all the works of an author, under whatever name they might be published,
should be brought together under a single heading.21 Not any more.
The title-page has become dominant, and the cataloguer, one assumes, is less
and less expected to read the work, more and more to look at it and describe
what he sees. Now valuable as objectivity may be, this seems to me a negation
of what people can do better than machines.
Soon
after that first trip to the United States in 1966, my colleague and I gave a
series of talks to our colleagues at the Museum about computers and how they
might be used in libraries, and especially how they might be used in the
British Museum Library. It was a disastrous exercise in public relations. We
were abused for selling out, for abandoning scholarship and scholarly
preoccupations. We were also asked what we saw as the role of people in our
machine-ruled Utopia.
We replied that the object of using computers
was to get librarians back on the reference desk, where their ability to read,
remember, and associate would be unchallenged by machine, and away from
repetitious and mechanical transcription and recording which were properly the
province of machines. This prospect did not go down well with the cataloguers
either.
I still think this is right, and I think it
is sad that progress, ISBD, MARC, and AACR, have been going determinedly in the
opposite direction.
There
is a basic question about cataloguing which has been asked more and more over
the past decade, and partly I am glad to say under economic pressure, and to
which no reply has yet been given which justifies our present standard
practice: what is the function of the catalogue? The question is fogged by the
assumption on the part of the providers of cataloguing that they therefore know
about catalogues.
It used to be said in contrasting the
cataloguing practice of the British National Bibliography and the British
Museum that the former, working without books for comparison or files to
maintain, catalogued each book as though it was the only book in the world,
while the latter catalogued each book in relation to all the other books that
had ever existed, whether it possessed them or not, Now both positions are a
little extreme, and both, it could be said, fail to afford any primacy of
interest to the non-librarian user. Unlike Panizzi, whose resolution of
difficult cataloguing decisions was generally achieved by invoking a pragmatic
utility to the user, which
is
why cataloguers find them so difficult to work with – there doesn’t seem to be
any theory behind it ...
One
of the examples there is the heading which is now abandoned called “Periodical
Publications”, which Panizzi had. Panizzi said “Well, you great institutions as
authors, so that if you have a periodical published by an institution then you
put it under the heading appropriate to the institution.”22
But then you will come across periodicals
which don’t appear to have authors, The New York Times, or something like that.
So what do you do with those? Well, you could scatter them throughout the
catalogue by title. But he said it would be simpler to bring them together in
one place, so that there was a large chunk called “Periodical Publications”
where you would have not merely those periodical publications without authors,
but also references to those which did have authors.23 This was not
really good enough. But it actually worked, and for certain types of
bibliographic enquiry, which had certainly not been foreseen by Panizzi, for
example the early bibliography of Norway or Denmark, by looking up
the
heading “Periodical Publications – Christiania”, you would find all the early
publications of Norway. But I’m afraid reason had to prevail, and it is now not
possible to discover in a single place the range of periodicals which the
British Library has.
One
of the functions of the catalogue in most libraries is to put the user in
contact with a book. Several people have advanced the notion that if this is
the prime function, then it can be performed with less cataloguing detail than
we customarily and standardly provide, that is to say, with short records.
Peter Lewis, the director of the British
Library’s Bibliographic Services Division, which produces the British National
Bibliography, is now veering in this direction, though not yet on paper.
Farewell logical and philosophical purity! He now finds that it is impossible
to contemplate a reduction from the seventeen weeks average which elapse
between the publication of a British work and the provision of the full MARC
record. So he is testing out the idea that he can improve currency if he can
persuade libraries that all they really need is information of the sort and
quantity which appear in the cataloguing-in-publication data. That plainly is
available at the time the book is published.
I
don’t think he’s yet seen that if that is so, there’s no need for a BNB at all,
because every library buying a book will have the cataloguing there.
But
if such a notion were to be both true and accepted, then the kind of cost
comparisons which have led to the widespread acceptance of OCLC might prove to
be the wrong ones, because it might still be cheaper for libraries not to look
up OCLC records, but simply to copy the CIP data, or the Library of Congress
data from the books which they have before them, and use lower paid, less
trained staff to do it.
There
may be other indicators in this direction. One of the aims of designers of
large central computer systems has always been the minimisation of storage, in
spite of the fact, clearly demonstrable at every stage over the past fifteen
years, that the unit cost of mass storage in computers is falling, and will continue
to fall in real terms, to the point now where I can buy a single hard disc to
be driven by a micro-processor costing less than 100,000 dollars which is
capable of holding a million Bodleian catalogue records – of course, we already
have short records.
This may point to a division of function of
our present catalogues: in-house short records, finding-lists, and
bibliographical databases available on such commercial services as Lockheed,
just like the abstracting services. You pay more for that.
There
are other pointers, still, to different ways of using electronic hardware.
There is now available a free-text searching engine, using at a key point
decidedly non-Von Neumann logic.24 And into this, you can simply
type catalogue records, the whole text of books if you wish, and you can search
on the strings. You will be able to type in a MARC record, tags and all, and
search on it as efficiently as typing in just the data with no tags, because
what you would be generally speaking looking for is the information and not the
structure.
There now exist browser terminals without
keyboards – people touch the screen – where users can be guided and can guide
themselves though catalogue searches. And there is much talk on the margins of
that branch of study known as artificial intelligence of expert systems in
which the expertise of a practitioner in a field such as medical or geological
diagnosis is put into question and answer form.
It seems to me only a matter of time before
someone tries to replace the reference desk personnel by their own encapsulated
expertise, and I shudder to think where the place of librarians will be then.
Notes
1
Book Arts Press lecture 116, 17 March 1983. This series of occasional public lectures
was inaugurated at Columbia University School of Library Service in 1972. It
continues under the auspices of the Rare Book School at the University of
Virginia as “Rare Book School lectures”. Lecture 500 will be delivered on 16
July 2007 by James Green of the Library Company of Philadephia.
2
John William Jolliffe (1929-1985) was Assistant Keeper in the British Museum’s
Department of Printed Books, 1955-1970, Keeper of Catalogues in the Bodleian
Library, 1970-1982, and Bodley’s Librarian, 1982-1985. Obituary by R.J. Roberts
in The Bodleian Library record 12:1(1985), 1-2. See also note 11 below.
3
Terry Belanger was then Assistant Dean, School of Library Service, Columbia
University, and Director of the Book Arts Press, the lecture’s sponsor. He is
currently University Professor, Honorary Curator of Special Collections, and
Director, Rare Book School, at the University of Virginia.
4
Augusta Ada, 1815-1852, daughter of the poet Byron. In 1835 she married William
King who later became the first Earl of Lovelace. She financed Babbage’s work,
and is credited with having written the world’s first computer program.
5
Her actual words were: “The Analytical Engine has no pretensions whatever to
originate anything. It can do whatever we know how to order it to perform. It
can follow analysis; but it has no power of anticipating any analytical
relations or truths. Its province is to assist us in making available what we
are already acquainted with.” Note G in L. F. Menabrea: Sketch of the
analytical engine invented by Charles Babbage … with notes upon the memoir by
the translator Ada Augusta, Countess of Lovelace (in Bibliothčque Universelle
de Genčve, 82, October 1842).
6
This library, with several other national collections, became the national
library under the terms of The British Library Act of 1972. The BL came into
operation from 1 July 1973.
7
In the 1980s, the audience would have taken the name “Snyder” to refer to Henry
Snyder, who had been appointed director of the ESTC in North America in 1978.
But in the 1960s, the name referred to Samuel Snyder, who was appointed
Information Systems Specialist at the Library of Congress in 1964. His task was
to draw together a number of different automation initiatives in North American
universities with the aim of producing a standard approach to the formation of
a national pool of authoritative bibliographical data. The LC’s “Office of the
Information Systems Specialist” had been renamed “Information Systems Office”
in 1965. See K.M. Spicher: the development of the MARC format, 78 (in
Cataloging and classification standards and rules, The Haworth Press Inc.,
1996, 75-90).
8
Henriette Davidson Avram is the creator of MARC. She was born in New York City
on 7 October 1919, and after working for seven years at the National Security
Agency as a computer programmer and data analyst, she joined the Library of
Congress in 1965, where she was given the task of designing and implementing
electronic cataloguing. The project was completed in 1968, and MARC became a US
national standard in 1971 and an international standard in 1973. During her 26
years of service in the Library of Congress, Avram was responsible for most
automation and networking functions. She retired in 1992, and died on 22 April
2006 at the age of 86. She is the author of MARC, its history and implications
(Library of Congress, 1975).
9
Barbara Evans Markuson was an assistant of Samuel Snyder in the “Office of the
Information Systems Specialist”.
10
COLA was the re-named “Clinic on Library Applications of Data Processing” whose
second meeting was held in April 1964 at the University of Illinois in Urbana.
“COLA […] became a discussion group within ISAD [Information Science and
Automation Division of the American Library Association] in 1970. Five years
later, the group changed its name to the Library Automation Discussion Group,
and in 1981 it merged with the MARC Users Discussion Group to become the
Library and Information Technology Discussion Group. This group lasted until
1984, when it was disbanded because of low attendance.”
http://www.lita.org/ala/lita/aboutlita/org/1st25years.cfm.
11
Alexander Matthieson Cain was Assistant Keeper in the British Museum’s
Department of Printed Books, 1954-1966. He and Jolliffe, “two of the most able
of the younger Assistant Keepers”, were set to work on the task of library
automation in 1964, and visited the U.S.A. in autumn 1965 to investigate the
progress that had been made there. Cain left for a post in the U.S.A. in autumn
1966, and Jolliffe completed their confidential Report on the feasibility of
using automatic data processing in the British Museum, principally in the
Department of Printed Books alone in April, 1967. See P.R. Harris: A history of
the British Museum Library, 1753-1973 (London: The British Library, 1998), 636.
According to Robin Alston, “ In the history of library automation [this report]
is crucially important: every bit as important as the discussions which took
place at the Library of Congress in 1964 regarding the creation of a computer tagging
system for producing computer-based catalogues, the practical result of which
was the development of MARC, now the universally accepted tagging code for
library catalogues.” See Robin Alston: The Eighteenth-century Short Title
Catalogue – a personal history to 1989 (www.r-alston.co.uk/estc.htm). 12 It has
to be remembered that Jolliffe is talking in 1983 of the situation in the
mid-1960s. At the latter time IBM was still using BCDIC (Binary Coded Decimal
Interchange Code), a 6-bit character set with 64 values and which allowed only
upper case versions of the alphabetic characters. This was enlarged in 1970 to
EBCDIC (Extended Binary Coded Decimal Interchange Code), an 8-bit set with 256
values
and
including lower as well as upper case letters. Both sets had inconvenient gaps
in numerical values between the three sequences A to I, J to R, and S to Z.
EBCDIC was the standard used by the widespread IBM 360 range and continued to
be used by the company even after ASCII (American Standard Code for Information
Interchange) had become ISO 646 in 1972. When this talk was given IBM had just
adopted ASCII for the IBM PC which then ushered in the modern world of personal
computers.
13
Pollard & Redgrave's Short-Title Catalogue. By this time, the STC revision
(by W. A. Jackson and F. S. Ferguson) was in progress, and the second volume
(I-Z) had been published in 1976; the first volume (A-H) was published in 1986,
and the index volume in 1991. The full title of the work is A short-title
catalogue of books printed in England, Scotland, & Ireland and of English
books printed abroad, 1475-1640 (London: Bibliographical Society).
14
From 1966 to 1977 Michael Gorman was successively Head of Cataloguing at the
British National Bibliography, a member of the British Library Planning
Secretariat, and Head of the Office of Bibliographic Standards in the British
Library. He is the first editor of Anglo-American cataloguing rules, second
edition (1978) and of the revision of that work (1988). He is the author of The
concise AACR2, 3rd edition (1999). He has been Dean of Library Services at the
Henry Madden Library, California State University Fresno since 1988.
15
See note 10.
16
GOST (....) is an acronym for gosudarstvennyy standart (............... ........),
which means “state standard.”
17
The “Helsinki Accords” were the “Final Act” of the Conference on Security and
Cooperation in Europe, held in Helsinki in December 1975. The principles it
formulated for relations between sovereign states did much to reduce the
tension of the Cold War.
18
A. B. B.
19
At the time of Jolliffe’s talk (in 1983) John Carson Rather (born 1920) was
Assistant Chief of the Descriptive Cataloging Division in the Library of
Congress. In the late 1960s he became the Specialist in the Technical Processes
Research Office. He became Chief around 1972, and was in that position when he
retired from the Library a few years later.
20
The International Conference on Cataloguing Principles, organized by IFLA, was
held in the Unesco Conference Building in Paris, 9-18 October 1961, under the
presidency of Sir Frank Francis, Director and Principal Librarian of the
British Museum. It was attended by delegations from 53 countries, and observers
from another 22. The outcome was a statement of 12 principles known as the
“Paris principles”, which have served as the basis for all national cataloguing
rules ever since, including simply tore the paper off the teletype, and with a
covering letter saying “Please supply these cards” sent it off to the British
National Bibliography.
21
The first major English-language cataloguing code was formulated by Sir Anthony
Panizzi for the British Museum catalogue. His ninety-one rules were “sanctioned
by the Trustees on the 13th of July 1839”, and underwent continual revision
until 1936. They were published in the introduction to his Catalogue of printed
books in the British Museum, volume I [no more published] (1841) in a section
entitled “Rules for the Compilation of the Catalogue” ([v]-ix). Jolliffe’s understanding
of Panizzi’s position does not seem to accord with rule XI, which states:
“Works of authors who change their name or add to it a second, after having
begun to publish under the first, to be entered under the first name, noticing
any alteration which may have subsequently taken place.”
22
Panizzi’s rule XXXIII states that when the author’s name does not appear “on
the title”, if the work concerns or is addressed to a named person, that name
should be taken as the heading. Rule XXXIV goes on to say: “When no such name
of a person appears, then that of any assembly, corporate body, society, board,
party, sect, or denomination appearing on the title to be preferred …”.
23
Rule LXXXI. The heading “Periodical Publications” was to embrace “reviews,
magazines, newspapers, journals, gazettes, annuals, and all works of a similar
nature, in whatever language and under whatever denomination they may be
published.”
24
This refers to the Memex Search Engine produced by Memex Information Engines
Ltd. Jolliffe had just heard (in January 1983) of its existence and obtained
information which was to lead to a Bodleian-based British Library R&D
evaluation project (SI/G/627): the report by Geoffrey Neate and Lou Burnard was
completed after Jolliffe’s death and was not published. Memex was based on
piping pre-tokenised compressed text as fast as hardware would allow through
the multiple parallel channels on a proprietary hardware board. The hardware
was loaded with the user’s query tokens and returned data addresses whenever
these were detected in the compressed data stream. Memex was used by news and
legal databases and by a telephone directory inquiries service.