Re: The Feynman Lectures on Physics Online Edition

[Michael A. Gottlieb]

Hi, Oliver.

sigh … the problem of incompatible font metrics.

I always try to use the same font (family and size) for maths as for
surrounding text. I also do this in LaTeX/PDF documents whenever
possible. IMHO, having a different font for mathematics is an ancient
legacy that predates Unicode and Unicode fonts and which I don't find
aesthetic.

The Feynman Lectures on Physics (FLP) was published in 1963-65, so it certainly predates Unicode fonts! The printed books have always used different fonts for mathematics and for body text. In figures they use yet other fonts for math and text. Another font is used for chapter titles and numbers, another font is used for the tables of contents. There are yet other fonts used on the title pages. Yet, in my opinion, FLP is a beautifully set book. It's a classic who's look and feel we wanted to preserve as much as possible. Rudi Pfeiffer, who created the LaTeX manuscript, took great pains to make this so -- even the page breaks fall in the same places as the original (despite changes of content)! 

I disagree with your notion that only one font should be used in a document (such as a book). Different fonts are useful for different things. Typography and bookmakings are arts that should not be overly restricted with arbitrary rules, and using only one font in a book is very restrictive, in my opinion.

The online edition of FLP, which was derived from the LaTeX manuscript, is designed to look as much as possible like the printed books, and I think that within the limitations of common browser support it does a pretty good job of doing that, though by no means perfect: the books are better set. A lot of people have complimented the looks of the online edition of FLP.  And your observations and comments have made it look even better, for which I am very grateful.

For web pages I find this even more important. Most/all of the MathJax
fonts are not well suited for low definition screens, which are still
very common.

I'm not sure if this is an issue for us. At least no one has complained about the MathJax renderings being unreadable. If you will please define (with some metrics) what you mean by "low definition screens" I can tell you exactly how many of our 2.2 million readers (to date) used them. (We use Google Analytics to track such information.)

See [2] for what I tried with MathJax to use the surrounding font in

math. However, the calculated margins of MathJax are obviously based on
different font metrics and some of the alignment is recognizably wrong
when I change fonts with CSS.

Just for fun I recreated your experiment on a page of the online edition of FLP. It looked terrible, as promised. ;->

 

Did you file a bug report regarding the Don'tMatchFontHeight feature?

No, but I found another bug yesterday, in the latest release of Chrome, which I did report (after figuring out a workaround... or so I thought. Today I find the bug still occurs on some of our pages).

Best regards,

Mike