nb@COYOTE (Neenie Billawala) (01/27/86)
Dear Rick Furuta,
Finally, some enlightenment on what happened to last year's
font tasting attempt!
The following file contains:
1) some background info on how this font tasting came about
2) results of the font tasting
3) identification of the samples used in the tasting
I understand you can put files in a public area. You may want to do
this for the 2) results and 3) identifications. A few more people
have expressed an interest in looking at the samples, and their
comments could then be added to such a file, if there is interest
for it. Then again, it may be quite some time before the next
response... The last person has had the samples since Sept 85.
Please use your discretion as to how much to send to the list.
Thank-you.
NB
[[Editor's note---thank you very much for sending this information
along. I'm sending it to the list without revision. --Rick]]
******************************************************************************
Last May in 1985 amid heated discussions on LaserLovers regarding the
"objective" print quality of various devices/fonts/formatters, Les Earnest
proposed conducting a "blind font tasting" to the Metafont for Lunch Bunch,
an informal open meeting held on Fridays (no longer) at Stanford and
hosted by Don Knuth.
Since each sample potentially presented a different combination of
device/font/formatter,the group decided on the subsequent guidelines
to try to achieve some standardization in the sampling.
1. The font to be tasted was to be "a Times Roman"
2. 3 sizes of it would be requested: 8pt, 10pt, 12pt
The same font may shine at one size and fall apart at another; the
thought was that these three typical sizes would give a good average
representation.
3. The sample consisted of 1) the same text set twice, justified and
flush left (or ragged right), 2) a listing of the upper and lower case
characters, numerals and some punctuation, and 3) some typical typographic
test patterns.
4. Plain text, troff, TeX, and Scribe files for the same test sample were
provided.
5. Samples were solicited over LaserLovers. All devices/resolutions were
welcome, though the request was for original output only,
i.e., not a photocopy of the same. Their identities were to remain
anonymous by ascribing a color to each.
6. The tasting was open to all and tasters could reveal their identities
or not, as they pleased.
7. The nature of the comments was left to the individual.
Some suggested guidelines were 1) "preference" and reasons for it
2) comments about weight/evenness/how the 3 sizes compared within a font/
legibility/readability/particular problems of characters or of the font/etc
8. Unedited comments by the tasters as well as the identities of the
samples would be sent/made available to LaserLovers.
Results:
***A total of 20 sets were received.
***13 sets of 3 pages were included in the tasting;
received versions of the Computer Modern fonts were not included.
***People found it useful to look at the lot at one time, so individual as
opposed to group tasting was adopted. Most of the tasters expressed
the opinion that this was "a very educational experience",
"worthwhile doing" ...
This method of individual judgements and handing the samples off from
one person to the next took longer than we expected. The shortest time
the samples were in any one person's hands for tasting was one day,
the longest 4 months so far. Not everyone who has expressed an interest
to look at the samples has actually had the chance yet. A number of
people looked at them, but for one reason or another were unable to send
in their results. Currently 5 taste results have been collected.
Ideally, a tasting might have been done within a time frame of 2-3 months,
the results and identifications sent out, and then perhaps another
might be done. However, it has been so long since I have received
anything new that I am sending off what was received.
***The reader can judge the results and/or the taster as s/he pleases
Neenie Billawala
Metafont/TeX Project
******************************************************************************
******* font tasting respondents:
"Cyclops"
"Ganesha"
John Hobby <JDH@SU-AI.ARPA>
Don Knuth <DEK@SU-AI.ARPA>
Alan Spragens <@SU-AI.ARPA:SPRAGENS%SLACVM.BITNET@Lindy>
********From:"Cyclops"
This collection of font samples shows a remarkable diversity of style.
Clearly some are not very close to the traditional Times Roman.
The following remarks indicate my assessment of their readability
regardless of style.
Key:
B = Too Bold,
H = Horizontally cramped (character spacing too tight),
F = Too light and fuzzy,
L = Too light,
O = OK (i.e. good),
U = Uneven stem weights,
V = Vertically cramped (ascenders and descenders intersect),
X = Too large,
Sample 8 pt. 10 pt. 12 pt. Notes
Red O O O
Blue L U O
Green H O O
Yellow O O O
White F F F (1)
Purple U B X (2)
Orange O O O
Brown L O O
Grey O O O
Black O O O
Gold O O O
Silver V V VX
Aqua O O O (3)
Notes:
(1) The white samples look like they were prnted on a mis-tuned
liquid toner machine (e.g. a Canon LBP-10).
(2) The purple samples showed a formatting problem: the text was
clipped on the right.
(3) Unfair competition! Obviously from a typesetter.
Based on a quick scan, I liked Yellow the best on readability grounds.
Sign me as "Cyclops."
********From: "Ganesha"
Now that I have found it, I've looked through the specimens.
Rather than rate them, I was trying to identify them.
I was not able to identify all of them, but have compiled
a few notes that you could include:
NOTES ON THE TASTED FONTS
These are some notes about features of the text that might aid
in identifying the source of the fonts. I tried to guess the sources
solely by analyzing the images, but I found that I could not do so
by a simple, short inspection. It would take me several hours,
a magnifier, and specimen books from the major equipment and font
vendors. By attempting the analysis, I was responding to Richard
Furuta's question (of some months ago) about how one could identify
fonts and printers from inspecting the output. Furuta's query came
from a statement by Brian Reid that he (Brian) could make such
identifications. The original guidelines of the font tasting required
that tasters who were able to identify the sources should disqualify
themselves. Since I have been able to identify some, though not all
of the fonts here, I have refrained from noting a preference for any.
RED: This is not Times Roman at all, but a kind of Century Schoolbook.
Note the characteristic "Modern" Q and R. Appears to be a white-writing
printer, possibly a Xerox machine, on which it might be named either
Century or "Classic".
BLUE: A characteristically narrow and light Times, possibly made
from an 18pt drawing size rather than 12 point. The fitting is tight
but irregular. Use of ligatures! Possibly a Xerox Dover.
GREEN: Very similar to BLUE, but not identical; no ligatures. I'm
tempted to say its the same machine, or at least the same font on a
similar machine as BLUE.
ORANGE: Almost certainly the work of the OCLC "Metafoundry". A unique
rendering of the lower loop of "g", the descender of "j", and many
other characters. The fitting has a characteristic irregularity,
seen in the a (loose on right), j (loose on left) and certain
other characters. This appears to be a write-black machine.
However, which black-write device I can't say for sure.
SILVER: Has a very odd 8 pt that seems only distantly related to
the 10 and 12 point sizes. The larger sizes show the same condensed
(narrow) look as BLUE and GREEN, but the image is much darker, as
from a write-black machine. The 8 pt puzzles me a good deal.
The 12 pt has an irregular fit not unrelated to BLUE and GREEN.
However, the overall width is greater on the SILVER fonts, as
though almost the same master designs had been used
for two different printers.
WHITE: An unusually foggy image quality, doubtless due to the print
engine rather than the font design, makes this one difficult to analyze.
It actually is very similar to the YELLOW specimen. My guess would be
that both are from the same fonts. Possibly WHITE on an engine needing
service, and YELLOW on a tuned engine, or the former on a write-white
engine (QMS?) and the latter on a write-black (Laserwriter?).
BROWN: The 12 pt has a characteristically light "o", and in the
10 pt the capitals are noticeably darker than the lower-case.
PURPLE: An odd, near total loss of serifs at 8 pt, and rather wide
fitting distinguish this specimen.
GRAY, GOLD, and BLACK really deserve some further scrutiny, and
perhaps reference to other specimens, to identify characteristic
features.
Perhaps when the sources are revealed for these specimens, the
beginnings of a key could be assembled. Though it might not
be worth the effort.
********From: John Hobby <JDH@SU-AI.ARPA>
Red: Word spacing seems a little tight at smaller sizes.
Characters appeared a little bleary at small sizes.
The typography is not very good: all the left quotes
came out as right quotes and there are some strange line breaks,
especially between the left quote and the g in `g' at 10 point.
Blue: Edges were much too bleary especially at small sizes.
The font is very condensed, but at least there is good contrast
between intercharacter spaces and interword spaces.
The differences between sizes seems especially large.
Green: Edges were also very bleary.
The font is very condensed, too much so for my taste.
Yellow: This is very bold, especially at small sizes.
The edges are comparitively crisp.
This is at least as compact as the `Green' sample, but I do not
find it objectionable.
Overall, my opinion of this sample is higher than average.
White: Extremely bleary with very poor contrast.
There may be some redeeming features, but they are very hard to
find in this sample.
Purple: This is bolder than the other samples.
The edges are fairly crisp (and this is important to me).
The appearing size is larger than for the other samples, especially
at small point sizes. (Measuring it, I find that the baseline
seperation really is about 7% larger.)
I notice that all the left single quotes are missing.
Orange: The edges are very crisp, but the `jaggies' are rather noticeable.
The appearing size is fairly large.
The weight is a little uneven in places, and there is something
really wrong with the `g'. Also the ascenders and descenders interfere
too much.
Brown: This is nice and crisp, but the 8 point is a little weak.
The thick/thin contrast is too extreme in places, and the
stroke weights are somewhat uneven.
Gray: Crisp but very bold, especially one of the 10 points.
The characters have a large appearing size, and the difference
between sizes does not appear as great as with some of the other
samples.
My overall opinion of this sample is not bad, but it may be a little
too bold.
Black: The contrast is very good.
This is not as condensed as many of the other samples.
Overall, I would say that this is one of the best samples.
Gold: The edges are fairly crisp.
The word spacing is a little too wide, especially in the justified
paragraph.
Silver: Crisp edges, but very bold.
The appearing size is a little larger than some of the other samples.
The intercharacter spacing is a little tight for my taste, but my
overall opinion s favorable.
Aqua: Good contrast and sharp edges.
This is also one of the best samples.
My only complaint is that I would prefer less thick/thin variation
in the capitals.
********From: Don Knuth <DEK@SU-AI.ARPA>
[This took 90 minutes, and I can't think of a way to speed it up!]
RED: rather nice 8pt (which doesn't look like Times); parens strangely vanish.
10pt specimen poorly printed (e.g. l's have different weights)
[strange line break after quote mark]
12pt is considerably darker in color, it seems
BLUE: this looks like Times. 8pt rather tight fit. Ligatures help.
GREEN: like BLUE without ligatures; slightly tighter wordspacing.
YELLOW: the three sizes all harmonious; rather dark, but people like that.
Strange `s' in 12pt. No ligatures, otherwise the fit looks good.
The 12pt size seems smaller than 12pt.
WHITE: Is this a washed-out Xerox copy of YELLOW?
PURPLE: Missing letters at the right! No open quote. `12pt' looks more like
13pt, but perhaps I'm confused by the small examples of YELLOW.
`B' and `F' are too dark in 12pt; `h' and `l' too dark in 8pt; etc.
Letter `f' too loose at right. Lots of other little problems like that.
ORANGE: 8pt is successful. This sample is the only one that has correct (?)
spacing in `A, B, C'.
10pt also satisfactory, perhaps over-kerned (e.g. in `we').
12pt has problems: too dark, the `S' is way too large, there's
trouble at lower right of `d' and `u', etc.
BROWN: `v', `w' seem to ride high. 12pt sample hard to understand, since
the same letter comes out looking very different in different places.
8pt and 10pt are reasonably nice except for lack of ligatures.
GRAY: Nicely worked so that ligatures aren't needed. This is more extended
than Times (except the 12pt), and bolder, but nicely fitted and
readable. The hyphen is too wide.
BROWN: Rather like GRAY but not as `black'! The `w' too dark in 8 and 10;
the `s' seems light in 12. In general not bad, but dominated by GRAY.
GOLD: 8/10 setting makes it harder to compare; 10pt and 12pt lack ragged
right. `w' floats in 10pt. The 12pt not too bad, but fit seems uneven.
SILVER: Jumpy x-line on 8pt; this face doesn't go with the other two.
`s' hangs low; `k' too narrow in 10 and 12.
AQUA: Very poor fitting. `k' too narrow. weights uneven. etc.
Most of the low-res samples look better than this hi-res one.
Overall, I like GRAY best, YELLOW a close second, although both are too dark
for the text of a book; fine for business letters.
********From: <SPRAGENS%SLACVM.BITNET@Lindy>
red- broken in small sizes; generally crude looking with
flat spots, fixations, globs and uneven appearing
baselines (overshot the overshoot?)
blue- broken in small sizes; hairlines drop out; somewhat
crowded looking (an old-fashioned look); coherent
across different point sizes
green- hairlines drop out but hold better that red & blue
fonts; crowded but smooth rhythm and even appearance
with general lack of fixations
yellow- well made font for the technology of the sample;
hairlines and serifs holding; even rhythm, no
fixations; dense enough without being too bold
white- lot of lines dropping out or getting broken edges
(which must be the printer, not the font...);
coherent through size range; few fixations
purple- solid hairlines; a bit bold for a roman text font;
solid and even, but with some globularization
orange- very pretty, elegant, balanced design; poor spacing
in text, however, where fixations also become apparent;
note how dot of "i" falls on tail of "y" in line above
for example
brown- too bold the vertical strokes compared to the
horizontals, angles, and curves, making for soporiphic
monotony in text
gray- very nice, my favorite [this is all the note I have--
it would have been better if I'd said why I liked it,
eh?]
gold- odd proportion with big X-height relative to cap height
like that New York Times font; makes it look too wide
spaced to me; I find it hard to read, but perhaps a
different line length, etc., would make it better
silver- crowded
aqua- too many fixations
That's all the farther I got. I hope my opinions are of some
value to you, although I'm hardly an educated observer. -Alan
******************************************************************************
********identification of sample sources
Label Typeface source Printer/Formatter
------ ----------------------------- ----------------------------------
Aqua Monotype 327 Times Roman Monotype Lasercomp, Mk1
Troff
Silver Times Roman, Canon LBPX
Xerox Parc/ETH Zurich Lara
Gold CG Times DEC LN03
CompuGraphic DECRITE
Gray Classic Canon Lbp-cx, Imagen
Interleaf Interleaf
Brown CG Times DEC LN01
Compugraphic Corporation TeX
Black Classic Dataproducts, NP-24, Interleaf RIP
Interleaf Interleaf
Orange Times Roman Imagen 8/300
Metafoundry TeX
Purple Times Roman Imagen 8/300
Xerox source files + homebrew Scribe
translator
White Times Roman QMS 1200A
Adobe Scribe
Yellow Times Roman Apple Laser Writer
Adobe Scribe
Green Times Roman Xerox Dover
SU/MIT/CMU Grant Fonts (Xerox) Scribe
Blue Times Roman Xerox Dover
SU/MIT/CMU Grant Fonts (Xerox) Troff + dcat
Red Classic Xerox 8046 Print Server
Xerox 8010 Star Information System 5.0