Critiquing the Perseus Project website
Jorn Barger
longstanding notion to critique [3] that site.
- the homepage has six levels of html TABLEs (30 seconds to render on
my slow old Mac). Happily, they offer a text-only alternate.
- the graphical version has three columns, which creates a
where-shall-I-look-first quandary that I hate. The eye is drawn by a
map that looks clickable but (amazingly) isn't. Under the map is a
timeline that makes sense when you know what it is, but is far from
self-explanatory.
- Under the map is a link labelled "A graph of the places and dates
mentioned in this collection" which-- even more amazingly-- is just an
explanation _about_ the maps and timelines, but offers no way to
explore them directly. So new visitors are pretty well assured of
suffering repeated frustration!?
But their mapping and timelining are potentially very useful tools,
once you've mastered them. They're even implemented transparently
enough that an expert can link to a customised map (eg places
mentioned in Hamlet):
http://www.perseus.org/cgi-bin/patlas? [remove whitespace
plot=tgn%2C1000066%3BDenmark%3B12 before clicking! see
&plot=perseus%2CLaertes%3BLAERTES%3B8 footnote [4]]
&plot=tgn%2C1000088%3BNorway%3B5
&plot=tgn%2C7012349%3BWittenberg%3B4
&plot=tgn%2C7003553%3BElsinore%3B2
&plot=perseus%2CRome%3BRome%3B1
&plot=tgn%2C1123294%3BDane%3B1
&redraw.x=1&redraw.y=1
I found this link in the lefthand column of a Hamlet etext, but very
poorly labelled ('sites' normally means 'websites', on the Web!):
Plot:
sites on this page
sites in this act <--- I chose this (Act One)
sites in this document
I'd rephrase this:
Show map of places:
mentioned on this page [etc]
The map it generates has serious auto-generated-SGML problems, at this
stage: 'Dane' and 'Laertes' aren't really places (and 'Rome' is
unimportant) so the map will just confuse students. Also, each of the
placenames is printed followed by an unexplained number, which turns
out to be the number of occurrences of that placename in the specified
text. And even countries are represented by dots that look like
cities (it would be nice if the dot could be made invisible).
But one _can_ hand-edit to url down to:
http://www.perseus.org/cgi-bin/patlas?
plot=tgn%2C1000066%3BDenmark%3B
&plot=tgn%2C1000088%3BNorway%3B
&plot=tgn%2C7012349%3BWittenberg%3B
&plot=tgn%2C7003553%3BElsinore%3B
&redraw.x=1&redraw.y=1
which removes the ringers and also the number-counts after those
"%3B"s (these could also be reformatted as "%3B(12)" if you prefer,
for clarity):
plot=tgn%2C1000066%3BDenmark%3B(12)
and I want to add some places from other acts, so I've clipped these
from those acts' maps:
&plot=tgn%2C7002445%3BEngland%3B
&plot=tgn%2C7006366%3BPoland%3B
&plot=tgn%2C7008038%3BParis%3B
which gives a nifty map... except that Paris is named twice, England
only half-named, and Norway almost unreadable against the background.
(I tried tweaking these 'tgn' numbers to see if I could relocate the
dots... but they just disappear completely.)
[I expect to continue this critique, but I'll break off for now]
[1] http://www.robotwisdom.com/web/hamlet/
[2] http://www.perseus.org/
[3] http://www.robotwisdom.com/sites/perseus.html
[4] http://www.robotwisdom.com/weblogs/urlhacking.html
--
Robot Wisdom Weblog: http://www.robotwisdom.com/ "If you worry that
reading the news online will rob you of the serendipity factor you get
with the newspaper, Jorn Barger solves the problem." --Dan Gillmor
Jorn Barger
As I was exploring Perseus's mapping tools, my browser crashed, so I
resumed via a different route:
http://www.perseus.org/startingPoints.html
Surprisingly, for all the junk on their homepage(s), this link isn't
directly available there-- it's buried on the page called 'Help
Library' (non-standard terminology). The Help-Library page recommends
(1st) the FAQ, and (2nd) a special help-page for the homepage. (Only
a gang of PhDs could create a homepage that needs a help-page-- in
fact, the homepage-helppage is so unreadable I may decide to create a
Perseus-homepage-helppage helppage!)
'Starting Points' is the 3rd main link on the Help-Library page,
confusingly enough. (This sort of design can _induce_
attention-deficit disorder, I fear.) The page begins:
The Perseus digital library provides a variety of
resources on the ancient world. Here are a few
jumping-off places which will help introduce you to
what's in Perseus: >The Historical Overview<, the
>primary texts<, the >Atlas<, and some sample >art &
archaeology< searches, along with the large site
plans. General >help< is also available.
I clicked on the first link there (Historical Overview), and was
disappointed to find it was a #-link that jumped me down an inch.
(Always label these with the "#" sign, please.)
Their first real startingpoint is blurbed thusly:
Overview of Greek Culture and History
>Tom Martin's extensive Overview of Archaic & Classical Greek
History< covers not only political history, but also cultural,
religious, and social aspects of ancient Greece. The overview
has been cross-linked to relevent resources in other parts of
Perseus, including images from art and archaeology, passages
in the texts, and Greek terms defined in our lexica. These
links are fully functional in the web version of the Overview,
with the exclusion of those 2.0 images which we do not have
permission to distribute freely to the general public.
For some reason, the last two sentences of this paragraph are forced
down to size "-1", making one feel a bit like Alice.
But I was thrilled by what the blurb seems to promise: a member of
their team has written a onepage synopsis of Greek history, with each
Perseus resource linked at the appropriate point(s)...?
Alas, PhDs cannot do things so simply. Instead we get a booklength
table-of-contents, and when we start at page one, we get... 'piebald'
hypertext, with random blue words linked to the search-engine
(non-standardly renamed 'lookup tool', for the sake of PhD-egos).
Even the first section-title on the page contains a random linked
word:
1. Introduction to the Historical Overview in >Perseus<
which of course delivers ZERO content about Perseus-the-project, but
tons of random links about Perseus-the-Greek-myth.
And scrolling down we find... Nielsen Was Here-- this unreadable,
unclickable book will be broken up into paragraph-length pages.
Yeesh.
And when a page is about Plato, _every_ occurrence of his name will be
blued, linking each time to the identical query. BRAIN! DEAD!!!!
There _is_ a way to turn off the piebalding, via the prominent
'Configure display' link. (For some reason-- complaints from users,
maybe-- this is even one of the first links on the homepage.)
The display-config page is another multicolumn, PhD-'optimised'
ADD-inducer. To turn off the blueing, 'Lookup Tool Links' should be
set to 'No'. (Design tip: the 'Reset' feature for forms is just
confusing when the main button is called 'Set...') (Another design
suggestion-- put little images of the 'before' and 'after' for each
setting, for newbies.)
Returning to the now-de-blued text page, an interesting-but-confusing
design-experiment is the 'analog' page-navigator just above the
body-text: "Your current position in the text is marked in red. Click
anywhere on the line to jump to another position." This again should
probably be for experts only. (There need to be a lot more 'configure'
options, with defaults all set to newbie-mode.)
Below this bluebar there are four navigation links: left and right
arrows (for next and prev page), 'table of contents' and then a
'Go to' textbox where you can type in chapter numbers, if you
magically happen to know them. It's not really obvious that the
right-arrow isn't associated with the 'ToC' link, either.
Before abandoning this path, I'll click the 'Plot... dates in this
document' link in the left margin (url trimmed slightly here):
http://www.perseus.org/cgi-bin/timevis?id=Perseus%3Atext%3A1999.04.0009
This 'timeline' view is a near approach to one I'd actually wished
for, for my Joyce research.
The horizontal access is dates in history, from 1200BC to 92AD. The
vertical is pages-into-the-book, with chapters marked by horizontal
blue lines (and truncated tiny-type titles). If you scroll right and
down you'll see tiny red dots for each mention of a specific year on a
given page, forming a downsloping line that runs out after Alexander
c300BC.
If you click on a dot, the default-settings just zoom you to a
closer-up version of the same graph (useful mainly when the dots are
densely clustered, I'd guess) but if you reset the radio-button to
"find date in document" you'll get the text itself, where the date is
mentioned.
[t.b.c.]
Jorn Barger
> The horizontal access is dates in history
Continuing my trek down the StartingPoints page:
http://www.perseus.org/startingPoints.html
Introduction to the Primary Texts
Read your favorite >Greek or Roman work in translation< from
our texts, which include >Homer's Odyssey<, >Aeschylus'
Agammemnon<, >Aristophanes' Lysistrata<, >Cicero's orations
against Catiline<, and >Vergil's Aeneid<.
The first link is not very descriptive, but the page it delivers is
useful (though autogenerated, and so non-optimised for newbies):
- it's titled "Perseus Table of Contents" but many pages share this
title (good-design should differentiate, especially in the titlebar)
- the first thing on the page is a form labelled "Change selected
items to: Collection:..." This is usually a bad idea for the top
of the page, since one hasn't even seen the page yet! The bottom
would be better. And the term 'collection' is librarian-jargon
that's not obvious to average surfers.
- next we get: "Greek and Roman Materials [Plot sites] [Plot dates]
Collection includes 47266 images. Primary and secondary sources for
the study of ancient Greece and Rome >(More about this collection...)<"
Those 'plot'-buttons are confusing and entirely useless for this ToC
(I think). And since the ToC is specifically limited to _texts_,
the item about images is just pointless bragging.
- under this is a blue-green-red-bar giving the number of words in
English-Greek-Latin for this 'collection'. Fancy but more confusing
than useful for most visitors, especially since the list that follows
includes only the English ones!
- finally now we get to the texts themselves. They're arranged
alphabetically, with no obvious option to sort by date (which would
be much more educational).
- since this set is filtered for English-only, the parenthetic
"(English)" each time seems like overkill. The 2nd set of
parentheses offers the site-standard abbreviation for the text (not
explained).
- The title of the work is blued, which is better than the whole
entry, but I'd prefer using the abbrev so the title is more readable.
- after 'Xenophon' is a second alphabetised list of "Secondary
Sources". These are useful, but hard-to-read in alphabetical
order-- sorted by subject would be much more useful.
- a few of the 2ndary (but none of the 1-ary) sources have
descriptive blurbs-- browsers would appreciate more of this, I'm
sure.
- when you click on a title, you get to dive into reading it,
which is usually the right choice. (Some reference works might
better default to their ToCs.)
- the five sample classics singled out on the StartingPoints page
are also happily linked to page one. (The blueing tends to run
these together a bit, so just blueing the titles would be better.)
[And here my browser died again, surely from overheating caused
by the site's html-junk...]
The next item on the StartingPoints page is:
The classical texts can also be browsed >in Greek< or >in
Latin<, and there are many >tools< available for studying
the texts.
Having figured out the English 'ToC' page, the Greek and Latin
variants are unsurprising. That 'tools' link is a whole nother
can o' fish, howmsever...
Although this page is also titled "Perseus Table of Contents" it's
really a very different animal. It's also linked prominently on
every page, after 'Configure display' and 'Help".
Its default 'Plot sites' button, for example, delivers a map with
_no_ named locations.
There are 16 tool-links, idiotically arranged in alphabetical
order. The next-to-last of these is labelled:
>Perseus Table of Contents.<
Use this tool to browse the contents of the Perseus Digital
Library.
'Table of Contents' is an infinitely unintuitive name for a
'tool'-- they might at least have called it 'the tableS of
contents tool'. Since the page it's linked from shares the same
name, it's rather horrifying when you click it and a ***750k***
monstrosity starts loading, listing every item in every
'collection'. (This is maximally unfriendly design.)
The item listed above it (alphabetically) is:
>Lookup Tool.<
This tool uses keywords and phrases to help you locate a subject
within the Perseus library, combining terms from the art and
archaeology catalogs, primary and secondary texts, encyclopedia,
image captions, and bibliography.
You might expect a big hairy page full of forms, but in fact
this looks like a pretty basic search-engine. (PhDs apparently
find the expression 'search-engine' too common!) And with its
default settings, it even returns the same results as the
anonymous little searchbox at the top of every page.
There's an 'advanced search' link that only adds two more
settings (for sortings)-- not too advanced! With all the blank
space at the bottom of these pages, you'd think they could
spare us some basic 'help' hints like whether the search is
'all' or 'any' or 'exact phrase', but instead there's a 'go to
hell^Hp' link (echoed by a separate 'what's all this then?'
link): http://www.perseus.org/Help/VorHelp.html
This PhD-designed 'help'-page starts... with the theory of metadata!
Then we get "In the upper righthand corner of every Perseus
is a text-entry box which you may use to invoke the Lookup
Tool." (Do they really call pages "perseuses"???) Anyone who's
found this page must have some degree of experience (and
stamina) so this babytalk is dumb.
And the next section swings all the way in the other direction,
to mind-numbing gradschool abstractionism:
The Collection Pulldown Menu
To limit your search to a single collection in the digital
library, select the collection's name from the pulldown menu
to the right of the search button. By default, the Lookup
Tool searches all the collections.
What's been left out is an explanation of what 'collections' are,
what they include, what the possible dangers of sticking with the
default, etc.
The answer to my boolean-search query, more or less:
only exact matches checkbox
Check this box if you only want to see results that contain all the
words you type in the text box; leave it unchecked if your results
can contain any of the words. Enabling exact matches generally
produces more restricted results but may leave out pertinent
material; it is disabled by default.
This _sounds_ like boolean 'and', but in fact my (original) search
for 'Caligula Neptune' returns zero results, whether it's checked
or not, so it can't be boolean 'or' in either state...???
('Caligula horse' without checking the box returns a few pages with
both, so I think the 'help' description is just wrong.)
[tbc]
Gil
Hi Jorn, robotwisdom rules.
I used to visit Perseus back in my Win98 days, but now using Linux I
can't configure the Greek text. I've tried to install the fonts from
tarballs but as a newbie I keep having problems.
Anybody know of a strategy a Linux neophyte can use to set up the
fonts. Please be verbose; don't assume I will know what you are
talking about. I'm using the Gnome desktop in RH71.
gjl
Gil
gjl
Jorn Barger
www.perseus.org/cgi-bin/perscoll?collection=Perseus:collection:PersInfo
> There are 16 tool-links, idiotically arranged in alphabetical
> order.
The problem here is that they've implemented a help-page-- which
should be optimised for newbies-- as a database-dump, which can't
easily be optimised at all.
Rather than alphabetical order, they need to be sorted from most-basic
to most-esoteric, and categorised by tool-type.
Of those 16 'tools', half are specialised language-tools that are
listed once for Latin and a second time for Greek: Vocabulary Tool;
Words in Context; Morphological Analysis; and Search English
definitions. A ninth 'tool' does headword-search for both languages
at once.
Then you have the 'Lookup Tool' which is just the search-form, and the
'English Index' which looks like the advanced searchform with an
unexplained added scrolling-textbox that arrives with random text
already supplied!??
The "Perseus Table of Contents" is a 750k variant on the current page,
as I mentioned before.
Which leaves the "Atlas Tool", the "London Atlas", the "Art &
Archaeology Browser", and the "Virtual Reality Interface".
The atlas tools are a noble experiment that currently suffer from the
'chaotic hypertext' syndrome (it's easy to go around in circles and
hard to know whether you've been somewhere before).
There seem to be separate, completely unintegrated databases for
London-only and the-world-in-general:
world: http://www.perseus.org/cgi-bin/patlas
London: http://www.perseus.org/cgi-bin/city-view.pl
These have a bewildering set of help-pages:
http://www.perseus.org/Help/mapfaq.html
http://www.perseus.org/Help/londonmapfaq.html
http://www.perseus.org/Help/perseusatlasfaq.html
http://www.perseus.org/Help/atlashelp.html
The first three are similar explanations of what the buttons do, the
last is a tutorial but doesn't remotely resemble what I see on my Mac.
(I was hoping I could construct a custom-view of London for each of
Iris Murdoch's novels, which are usually very definitely situated in
that city's real geography, but several hours of wandering in circles
convinced me this is still impossible.)
The Art & Archaeology Browser is usable, but demands you step thru
many layers of 'thin' menus (only-one-word descriptions as links),
with few hints of what you'll really get for each choice. It should
probably just be called the 'image-browser'.
The Virtual Reality Interface is unbearable on my platform, I think
because they use a 1-pixel background image which re-freezes my Mac
with every tiny window-scroll. For the little I could see, it looked
pretty confusing, but great potential (eg QuickTime panoramas). The
name is poor, because the interface itself isn't VR, and most of the
stuff it interfaces to isn't either.
In order to squeeze the url of the 'tools' page into 80 columns, I
trimmed a bit off the end:
www.perseus.org/cgi-bin/perscoll?collection=Perseus:collection:PersInfo
&type=interactive+resource
^^^^^^^^^^^^^^^^^^^^^^^^^^
which allowed a helpful list of help-pages to appear after the 16
'tool' links.
This list surely overlaps the "Help Library" page:
http://www.perseus.org/Help/
...but because the urls are often linked via indirection, you can't
reliably use your browser's visited-link color to determine which
you've seen.
Jukka K. Korpela
> I wrote, re:
Yes, we can see that you are talking to yourself, but why do you that
in public, and in three groups?
"7. If a message has been sent to more than one group, the odds are that
none of the groups is the right one.
42. The average usefulness of a thread is inversely proportional to the
cube of the number of groups it is posted to."
http://www.cs.tut.fi/~jkorpela/usenet/laws.html
Followups randomized as usual.
--
Yucca, http://www.cs.tut.fi/~jkorpela/
Jorn Barger
www.perseus.org//cgi-bin/perscoll?collection=Perseus:collection:PersInfo
(which I'll call the ToC page)
and to compare these to the main 'Help Library' (HLib):
http://www.perseus.org/Help/
The last item on ToC is called "Using the Perseus table of contents"
so you'd think it would be prominently offered at the top of each
toc-page... but instead toc-pages offer a generic link to HLib,
where the toc-help link is buried 12 links down.
http://www.perseus.org/Help/tochelp.html
Perversely, this page uses the singular "the Perseus table of contents"
to refer to any/all the pages generated from the toc-database.
I don't see any warning that the full ToC is 750k-and-growing.
Design suggestion: the 'Change selected items' form at the top of every
toc-page should be reformatted to be self-explanatory. Currently it says:
Change selected items to:
Collection: Type: Language:
[Perseus Tools and Information] [Any] [Any]
Keyword(s): [ ] [Change]
It might better read:
Below is a listing of all resources in the
[Perseus Tools and Information] collection, filtered to include
only resources of type [Any] written in language [Any]
and with the descriptive keyword(s) [ ]
(Note: keyword-search does not dive into the full-text, just
the 'metadata' in the indexing-headers.)
You can customise the listing by resetting any of these values
and clicking this [Change] button.
Warning: the full ToC (the 'Any' collection) is very large.
A rough picture of the size of the collections:
nsdl-alist: one screen
Perseus Tools and Information: 6
Perseus Project Research Preprints: 7
Greek and Roman Materials: 63
Renaissance Materials: 15
Archimedes: Sources for the History of Mechanics: 1
The Bolles Collection on the History of London: 8
American Civil War: 11
American Memory: California: 136
American Memory: Upper Midwest: 153
American Memory: Chesapeake Bay: 25
Tufts University History: 2
Duke Databank of Documentary Papyri: 35
Beazley Archive: 1
Boyle Work Diaries: 8
Objects under development:1
(It's a bit tricky to count these because there are subheads for
each collection that are formatted more prominently than the main
heads!?)
These screen-counts for the American-Memory collections is misleading
because each resource-listing includes a lengthy description-- most
collections include no descriptions at all.
The ordering of the collections is opaque-- roughly-chronological
might be better, with the meta-collections (eg tools) at the
beginning.
What the site as a whole really needs is the same thing I recommended
for the LoC's American Memory site [1]-- a single page that briefly
explains each collection, in order of popularity, with direct links
to the best resources and samples that make clear what sort of content
is there.
The rest of the toc-help page is written in mind-numbing prose with
no 'live' examples. (Hello??? Instead of pictures of searchforms,
why not include working examples?)
Returning to the list of help-files on the tools-toc, a lot of these
are specific to particular tools: atlases (discussed briefly in my
last episode), timelines, text-and-word-study tools, etc. There's
no reason except database-dump-clumsiness that these shouldn't be
listed together with the relevant tool-links.
[Oops-- that 'text-and-word-study' help page turns out to be the
most useful helppage on the whole site, though you'd never guess from
the title-- it tells how to read the actual texts. The blurb reads
"Text and Word Study Tools Help. A guide to using Perseus texts and
word study tools. (English)" So only if you read carefully will you
notice it offers texts-help, and not just text-tools help.]
Two meta-meta files discuss policies and copyrights. One that seems
not to belong in this list claims that it "describes papyrus
resources" but actually briefly lists the papyri-relevant tools.
Greek-font help also seems like a misfit here.
And there's a page comparing the CD-version with the website itself
(though the website is misleadingly identified only as the "Perseus
digital library").
And there's the webmaster's faq (faq.html should be called
wwwfaq.html) and the lookup-tool faq.
And finally the first link is "About Perseus. General information
about the Perseus Project. (English)" which doesn't mention
collections at all.
Comparing that tools-ToC help-texts list to the HelpLib list, the
HLib favors the wwwfaq first. Then it offers homepage help and
StartingPoints, neither of which were in the ToC list at all.
It offers a 2nd Lookup-Tool tutorial page which wasn't listed in
ToC. And a Vocabulary Tool helppage that wasn't there. And one
for the VR collection. And a links-page for CDs-info.
The next section of the HLib is six overviews of specific
collections (though the toc-form lists 16 collections, and the
headers at the top of every page list either 7 or 9). The first
is for the Classics collection:
http://www.perseus.org/cgi-bin/ptext?doc=1999.04.0053
That url is geeky, but may give a clue how out-of-date the overview
is. (The page itself is undated.)
The overview is short, written in the standard mind-numbing academic
prose. (The word 'clasics' is even misspelled at one point.) There's
no sense that the authors feel any real enthusiasm for the site, or
even that they've ever used it personally. This may indicate that
everyone knows it has serious problems, but can't be honest about
them for 'political' reasons. (At Northwestern they used the
expression 'career-limiting move' or CLM.)
With the lookup-tool links turned off (via configure-display) it has
only seven live links, to four main reference works that have been
transferred from paper to Perseus, then the booklength historical
overview written for the site, and finally two sample-links to
Julius-Caesar etexts.
Newbies by default have the lookup-links turned on, and get dozens
of showy-but-useless links for random words. (Eg 'secondary sources'
is especially bogus-- it gives a lookup-tool dump whose first item
_appears_ to be a custom 'Perseus Encyclopedia' article on the
difference between 1-ary and 2ndary sources, but in fact is a random
article on Plato that happens to highlight those terms. What's this
'Encyclopedia' anyway, and why haven't any of the overviews mentioned
it yet?)
I have to turn the lookup-links back off to find the texts remotely
readable. And because the terms are so unintuitive, using that
configure-display page always causes great anxiety.
In the upper left are two links labelled:
Classics
>Collection contents<
>About the collection<
The first links to a standard ToC-dump, the 2nd to... this same
article! But the url is disguised so the link won't be colored
purple:
regular: /cgi-bin/ptext?doc=1999.04.0053
disguised: /cgi-bin/ptext?doc=Perseus%3Atext%3A1999.04.0053
(%3A is a colon, %3B a semicolon, %3D an equals-sign, %2C a comma.
These can be safely substituted if you're trying to shrink a url
down under 80 chars, I'm pretty sure.)
Under these two are three more:
>Greek Hist. Overview<
>Art & Arch. Catalogs<
>Other Tools & Lexica<
The first is the booklength history, the second is a very short
toc-dump, filtered on the keywords "Perseus catalog", the 3rd is
our good friend the tools-ToC.
(I'm noticing a terminological inconsistency between 'Classics'
and 'Greek and Roman' that's bound to confuse newbies. Both terms
should be used wherever space allows: "Classics (Greek and
Roman/Latin)".)
The Renaissance Collection Overview:
http://www.perseus.org/cgi-bin/ptext?doc=1999.03.0024
is dated April 2001 at the bottom, so probably Mar1999 was when
it was created (?) and this author is a little less disengaged
from the project.
He consistently uses an artsy title (The Perseus Garner) for the
English renaissance colection as a whole, which I've seen used
nowhere else-- it would just add confusion. The first sentence
describes the 'Garner' as "the latest collection" which is obviously
wrong-- web-authors have to learn to avoid such constructions. The
overview is a bit dated but generally tries to give the right sort
of information. There's a big hyperlinked table of
Shakespeare-editions that could have been handled a little more
elegantly. Eg a TABLE-free list:
The Comedy of Errors: >Folio<, >Globe<, no Variorum
The lefthand column adds some new menus, first "Dictionaries".
This links to a search-form that turns out to be an as-yet-unhinted
variant on the Greek-Latin-headwords tool (aka "Dictionary Entry
Lookup"). There's no hint what dictionaries it will access-- my
test search for 'the' dropped me into an impressive entry for
Shakespearean useage of 'the'. My test-search for 'bugger' came
up empty-- it just gives a new-search link (why not an explanation,
and a new blank form???).
There's a link "Click here to search English definitions in the
English Lexicon." which (unintuitively) turns out to mean 'search
text within definitions' (instead of just keywords). No clue what
'English lexicon' is jargon for.
Under "Dictionaries" comes:
Plot:
sites in this document
dates in this document
(No warning that these open in a new window.)
'Plot sites' is always good for a laugh-- here the worldmap
includes Othello, Delaware; Macbeth, South Carolina; Verona,
Minnesota; Venus, Tennessee; and Marlowe, Indiana.
'Plot dates' finds 1577, 1623, 1808, and 1891, but misses 1604,
1616, and 1909, clearly because they're 'hidden' as anchortext
(ie blued).
Under "Plot..." comes "Images in this document" which _doesn't_
launch a new window. It links to an image-browser with the single
thumbnail of WS, and clicking on that gives you a bigger version
(which proves to be a drawing not a photo, but no explanation of
why or who drew it). Strangely, the thumbnail within the overview
itself is blue-bordered and links _directly_ to the same page, but
_does_ launch a new window.
Finally in the lefthand column we get:
Contents:
The Works of William Shakespeare
The Works of Christopher Marlowe
Primary Sources
Secondary Sources
The Texts
which, stupefyingly, are _not_ just undeclared #-links as you
might think-- they're exceedingly-braindead links back to the (top
of) the same page itself!?!?
HELLLLLLLLP!!!!
[tbc]
[1] LoC-AM bashing: http://groups.google.com/groups?th=dedc61d460737c79
Jorn Barger
the Bolles Collection on London:
http://www.perseus.org/cgi-bin/ptext?doc=2000.01.0043
It features a thumbnail photo of Prof Bolles taken, apparently,
during a total eclipse. The first paragraph reveals the new info
that this collection is actually about _Victorian_ London (which
would have been nice to mention in all the standard links).
The second par is a sales-blurb summarising the collection... but
the rest of the 'overview' is a bio of Bolles, so this one gets
an F for utility. (Sales-blurbs do not belong on help-pages. The
opposite is what's needed-- a ruthless inventory of where the
collection fails to live up to its ideals.)
The single new marginal link is called "London maps" and links to
the London-Atlas-tool's default page, which offers zero orientation,
only a link to londonmapfaq.html, which itself is just tool-tips,
with no hints about what content is available or how it can be
used.
Peeking at the ToC for this collection, I notice some Dickens and
Defoe that should be popular. Clicking on 'Bleak House' to see what
kind of hyperlinks have been added, I get a Preface with no visible
links or notes.
(It would be nice if, instead of having to go back to the configure-
display page to make sure I've got links turned on, there were some
links at the bottom or on the side that said "Click here to turn on
notes from [Source]". But they are turned on, and there's just no
notes yet.)
The main text is labelled "[p. 5]" (no indication of what edition,
though). Apparently they've chosen to replicate the paper pagination
on the website, to the extent of _breaking sentences_ in the middle.
Clicking the (unlabelled) next-page arrow changes this semi-sensible
url:
/cgi-bin/ptext?doc=Perseus%3Atext%3A2000.01.0032
to:
/cgi-bin/ptext?doc=Perseus%3Atext%3A2000.01.0032
&layout=
&loc=5
&query=page%3D%232
This delivers page six, whose prev-page button links to:
/cgi-bin/ptext?doc=Perseus%3Atext%3A2000.01.0032
&query=page%3D%231
&layout=
&loc=6
Digging out my HTML-chart I find that "%3D%23" converts to "#="
so 'loc' is the _paper_ page# of the webpage we're on, and 'query'
is the _web_ page# for the page we want to go to: "page#=1" to
return to the first page (aka p5). Since both '=" and '#' are
special characters in this context, you *can't* safely convert
them, alas. (They really should have avoided that encoding!)
Trimming away (by hand) either the 'query' or the 'loc' still
delivers the correct page (and of course the 'layout' is just
html-junk here).
In fact, you can even get it down to:
/cgi-bin/ptext?doc=2000.01.0032&loc=6
by ruthlessly hacking out that "Perseus%3Atext%3A".
Jumping ahead to "Installment 1" by clicking the marginal ToC, we get
a new url-element:
/cgi-bin/ptext?doc=Perseus%3Atext%3A2000.01.0032
&query=head%3D%232
ie "head#=2" for heading-number. Even more horrifying, the page
displayed is paper-page 7, which includes no more of installment one
than the mere heading alone.
And still no annotations. (For laffs I asked it to plot-sites, and
got an empty worldmap-- would it be so hard to suppress that link
until there's something to make fun of?!?) Defoe is the same deal.
Next overview: "American Memory: California":
http://www.perseus.org/cgi-bin/ptext?doc=2000.02.0207
I'm noticing a pattern here (at the risk of violating the DMCA on
reverse-engineering):
1999.01: Greek and Roman Materials (1-ary only)
1999.02: ???
1999.03: Renaissance Materials (1-ary and 2-ary)
1999.04: Greek and Roman (2ndary materials only)
1999.05: Duke Databank of Documentary Papyri
1999.06: Archimedes: Sources for the History of Mechanics
2000.01: The Bolles Collection on the History of London
2000.02: American Memory: California
2000.03: American Memory: Upper Midwest
2000.04: Tufts University History
2000.05: Perseus Tools and Information
2000.06: Perseus Project Research Preprints
2001.01: American Memory: Chesapeake Bay
2001.02: ???
2001.03: ???
2001.04: ???
2001.05: American Civil War
2001.06: ???
2001.07: Boyle Work Diaries
2002.01: nsdl-alist
???: Beazley Archive
So they've enhanced the geekitude of most urls by using an inflated,
essentially-meaningless, utterly-non-mnemonic internal code for each
collection... Thanks, guys! (NB: leading zeroes are non-optional.)
The California overview is three webpages, no annotations, in standard
size=-1 font. The atlas-links may actually be somewhat useful. Two
new marginal links labelled ">American Memory< Courtesy of the >Library
of Congress<" lead 1st, a bit confusingly, to the LoC's separate 'AM:
California' section, and 2nd (uselessly) to the main LoC homepage.
Also, next to the flag/book icon there's a dup link to the
collection-toc.
AM: Midwest looks unremarkable.
The Tufts-History overview is at:
http://www.perseus.org/cgi-bin/ptext?doc=2000.04.0006
Someone has put their foot down and demanded the Perseus-straitjacket be
loosened here. The 'sites in this document' link remains, much more
prominently placed, and links to a vast panorama of the western USA,
with poor little Medford, all alone, trying to sneak off, stage right.
The 'dates in this document' link fares worse-- 3 dates in four
paragraphs (1852, 1852, and 1950). It finds the '1950' in an image
caption but misses 1993 in the last par, for some reason.
Whoever did this facelift-- probably Tufts' pr dept-- had at least a
reasonable sense of how to communicate via the Web.
HelpLib skips the other six-odd collections' overviews, and jumps to an
overview of research preprints, which warns that some are offsite and
some are pdf, but says little about their content or highpoints. The
ToC is arranged in random order and doesn't hint which are pdf. One
of the links is back to the overview (?!?) and another is to a
bibliography which it falsely claims it's in chronological order. (This
same bibliography is also linked at the top of most/all pages.)
Before looking for other collection-overviews, the last few links in
HelpLib are "General and Contact Information":
"News & Announcements" is a useful timeline of press releases from
1995 to the present. "About Perseus" just adds to the fragmentation
of the helpfiles. "Copyright & Permissions" is unremarkable, except
for the unnecessary three-column design for the addresses-list.
"Policies of this Site" is miscellaneous dated trivia. "Support
Perseus" invites $1000-donors!? "Contact Information" is pretty
uncontroversial.
Boyle Collection overview (notebooks of 17thC chemist):
http://www.perseus.org/cgi-bin/ptext?doc=2001.07.0047
Amazingly, the ToC is undescriptive, but links to a 'toc' document
that _is_ descriptive... but that one lacks the hyperlinks!
Papyrus Collection overview (just tools):
http://www.perseus.org/Texts/papyri.html
American Memory: Chesapeake Bay has a basic, linkless overview:
http://www.perseus.org/cgi-bin/ptext?doc=2001.01.0142
I don't find any overview page for the Archimedes Collection (history
of mechanics). This is the url they give for the ToC:
http://www.perseus.org/cgi-bin/perscoll?
collection=Perseus%3Acollection%3AMPI
&type=Any
&lang=Any
&lookup=
&.submit=Change
This can be trimmed all the way back to:
http://www.perseus.org/cgi-bin/perscoll?collection=MPI
Nothing for the Civil War but a ToC:
http://www.perseus.org/cgi-bin/perscoll?collection=cwar
though it includes a 1901 encyclopedia of US history:
http://www.perseus.org/cgi-bin/ptext?doc=2001.05.0132
whose main page (alas) is 140k of front-matter, and whose navigation
gets worse from there...
The Beazley Archive's ToC has one item, which leads indirectly to
their own 'pottery database' website. Here's an above-average url:
http://www.beazley.ox.ac.uk/BeazleyAdmin/Script2/Pottery.htm
[almost done]
Edwin Menes
Perseus itself. Not many working Classicists check in here regularly.
Second, Perseus originally (before there was a website) had more to do
with sites, art, architecture, artifacts than with texts. Texts
(originals, translations, dictionaries, etc.) have been added as
copyright and labor have allowed. The architecture of the site may well
be more friendly to some kinds of information than to others.
Third, Perseus is a work in progress. It is not surprising that parts
of it do not fit well with other part at this point. I have no idea
what plans for the future may entail.
Conclusion: tell Greg Crane at Perseus what's on your mind, not us. He
can see to it that things get fixed as the site develops. We can't.