Quote: Everything should be as simple as it can be, but not simpler (antedating attrib Albert Einstein 1950)

victor steinbok aardvark66 at GMAIL.COM
Thu Feb 25 19:39:05 UTC 2010

Responses are interlaced with quoted text.

On Thu, Feb 25, 2010 at 2:04 PM, Garson O'Toole
<adsgarsonotoole at gmail.com> wrote:

> Thanks for your response Victor. I think the snippet you located that
> Google Books dates as 1949 is really a snippet in the June 1950 issue
> of Poetry that I have verified on paper.

You are undoubtedly correct. It is quite common for journals/magazines
to start volumes in mid-year and I did not check the practice of this
one. And Google date tags are often mistaken (for example, a book that
appeared in 1964 has two entries--one for 1964 and one for 1963).
However, this does not address the other point. The 1981 edition of
the essay has the date 1948 at the end of the first part (it has 1928
at the end of the second part). These may be nothing or them may
indicate the actual dates of composition. Not knowing much about the
essay itself or the author, I can't speak to the meaning, I can only
suggest it as a subject for investigation. What is helpful, however,
is that, unlike other editions, the 1981 edition of Prepositions is
available in limited view, particularly with the text addressed to WCW
visible in full.

> Before verifying cites on
> paper I typically use "date probes" to check for plausibility when
> document viewing is restricted to snippets. In this case probing the
> document with dates provides very useful information.
> I am certain that the following paragraphs describe an obvious
> strategy to you Victor and to many other list members. But I include
> them here because they may be useful to other individuals who are
> starting to use Google Books for research and who are confronted with
> the aggravating restrictions imposed by snippet view combined with
> notoriously inaccurate metadata.

I generally use this trick for stubborn or suspicious data. For
example, in another thread, I cited the second volume of an
encyclopedia that has 1811 clearly printed in Roman numerals on the
title page. However, the other two volumes are dated 1835 and 1837--in
all copies on GB and are also clearly dated the same way on the
respective title pages. Simply searching for a succession of
four-digit strings 1819-1841, I managed to find internal references to
actual events--as having occurred in the past--for virtually every
year from 1819 to 1838 and none for 1839-1841 (and a couple of other
years). This clearly placed the volume in the same general range as
the other two. Subjectively, I estimated the publication date to have
been 1841 because that would require an omission of only a single
character to appear as 1811. Matching "related" dates also works well
in a more direct manner when resolving the volumes of journals--if
other scanned volumes match the date-volume correspondence, it
enhances the credibility of the cited data. Unfortunately, a large
fraction of GB-scanned journals are dated to the "founding" dates or
other dates that may appear on the cover (e.g., if the title page of
the first issue in 1982 has "1984?" in large font, GB may interpret
that as something published in 1984).

As I mentioned, I also used a similar approach in dating the 1954
article in Forbes that I included in this thread. I first verified
that the volume numbers matched other Forbes volumes (and data found
in WorldCat). I then searched for "1954" and a few subsequent years.
The snippets that appeared for "1954" pointed to financial events
(stockholder meetings/votes, maturity dates, etc.) that are expected
to occur at specific dates in 1954, as if it is the future. I found no
hits for a selection of subsequent years. So the 1954 GB dating of the
volume appears accurate (although it is not clear which issue the
snippet appears in, but this can be estimated through other means). I
also mentioned earlier that the illustration that appears on the page
looks like something that might have been available and advertised in
1954 (plug-in timers are no longer advertised in Forbes AFAIK).

Other means of ascertaining the dates include external citations to
the title in question (does not always work well for journals and
collections, where it is not clear which individual piece is being
viewed, or for obscure works that are not commonly cited or sometimes
cited incorrectly). Finally, the third, although less reliable method
is to check the WorldCat metadata and specifically to track down the
volume in question at several libraries. This is particularly useful
for matching journal volumes to dates, although this approach in no
way can resolve the problem of the content not matching the entire set
of GB metadata.

In any case, thank you for the clear articulation of the method.


The American Dialect Society - http://www.americandialect.org

More information about the Ads-l mailing list