New to Typophile? Accounts are free, and easy to set up.

Posted by Gus Winterbottom in

16 Jun 2011 — 5:37pm

16 Jun 2011 — 5:37pm

Are there any readily available fonts that have an upper-case letter R with single dot above, and an upper-case letter R with a double dot (or diaeresis) above? These symbols are used in a technical discussion of radar; I think they pertain to range-rate.

16 Jun 2011 — 7:16pm

Readily available? I doubt it. It's (to my knowledge) not part of any standard spec. If you're typesetting this you could somewhat simply produce these as /R/ with two different character styles applied.

16 Jun 2011 — 7:04pm

You could contact Sindre Bremnes about his charming Telefon.

16 Jun 2011 — 7:32pm

Some fonts have R dotaccent (Unicode 1E58/1E59) but I've never seen any with R dieresis.

Use combining accents instead.

Copy & paste the following into a word processor and it'll work with some system fonts such as Corbel, Consolas, Calibri, Cambria, Candara, Deja Vu, Segoe, Times New Roman.

R̈r̈Ṙṙ

16 Jun 2011 — 8:56pm

Thanks all. I finally figured out how to do it with Word's equation (EQ) field code. Finding the right diacritic marks was the hard part.

16 Jun 2011 — 8:59pm

You could also look at the SIL fonts such as Gentium and Andika

http://scripts.sil.org/cms/scripts/page.php?&cat_id=FontDownloads

http://scripts.sil.org/cms/scripts/page.php?&item_id=andika

http://scripts.sil.org/cms/scripts/page.php?&item_id=Gentium

I checked the R with dot above ( uni1E58 )in Andika. It has it. R+Dieresis - no.

David and Frode's solutions sounds the best to me.

17 Jun 2011 — 3:52am

Assuming your Word file is the final version, then the EQ field is fine.

But if you're subsequently sending it to a publisher/typesetter who are going to import it into some other application, it might break. Either the combining accents, or a regular spacing accent kerned back over the R (which you can still also do in Word) should at least survive the transfer into another system (and the latter would be less dependent on picking a font with a large character set).

17 Jun 2011 — 12:03pm

Ray, how did you construct those characters? I get inconsistent results when I paste them into Word. Depending on what font I apply, I can sometimes see the double dots above the first two Rs, but the third and fourth Rs are always sans and always have a square dot no matter which font I apply. However, they do make Word show symbols from the Latin Extended Additional area that normally don't show up there. They also make Thai show up in the Symbol list when it normally isn't there.

Re DTW's suggestion to use combining accents or spacing accents, those don't seem to work (outside the EQ field code) unless Word has non-English or right-to-left language support installed. But I may be doing something wrong.

17 Jun 2011 — 3:28pm

To avoid headaches maybe you should commission

these two additions to the font(s) you're using?

If there's a budget: hpapazian at gmail dot com

hhp

17 Jun 2011 — 7:21pm

I think something must have happened to the encoding when I posted it. Try making your own. Combining diacriticals are accents with zero width and a left offset. If the font doesn't have proper combining diacriticals, this trick won't work . . . which is why your should stick to Corbel, Consolas, Calibri, Cambria, Candara, Deja Vu, Segoe or Times New Roman.

I used OpenOffice to construct them. After each R, I used the insert symbol command, chose "combining diacriticals" and selected the appropriate accent.

18 Jun 2011 — 11:26am

If your dotted R is a rate of change, then your text is using Newtons's notation. The dot and double dot are then mathematical accents and their centering differs from that of text accents. Here is an example taken from Vieth, Math typesetting in TEX: The good, the bad, the ugly. On the right is the V with the mathematical accent circumflex.

Moreover, if R with a dot above is a mathematical formula, then the R should normally be in italics except in France.

18 Jun 2011 — 11:29am

I have to say that this "proper" horizontal placement issue seems pedantic to me.

hhp

18 Jun 2011 — 12:00pm

Well, those "accents" are operators, they are not semantically a component of the letter (at least in this particular case).

18 Jun 2011 — 12:03pm

So you're saying they're supposed to look detached? If so, a slight shift is

imperceptible, hence only really visible to the author! If the separation

must be explicit I would argue for placement like an apostrophe's.

hhp

18 Jun 2011 — 12:51pm

I am just saying there is no need to make them look like they are attached. Their position in mathematical fonts is determined by the "Top Accent Attachment" (see the Math Fontforge doc). I am just a user of mathematical fonts, with LaTeX. I never implemented those things, not even for LaTeX. You guessed correctly, derivatives are much more often denoted with apostrophes than with dots. The dot notation is used in some parts of mechanics, in particular in the so called Hamilton equations where the dotted x is semantically a variable and then the argumentation would get quite tricky.

18 Jun 2011 — 12:59pm

> there is no need to make them look like they are attached.

But is there a need to make them look like they are

detatched?If so, then a slight shift is pointless; and if not, then the "text accent" is fine.

BTW, just to be clear: I didn't mean to use an apostrophe; I meant

to position any explicitly-detached qualifier beyond the right edge.

hhp

18 Jun 2011 — 1:28pm

No, there is no effort to have them look detached, and they are top accents. Here is what I get with the

`\dot{x}, \dot{p}, \dot{q}, \dot{r}, \dot{R}`

in LaTeX with the Computer Modern font:I must confess that, to my eye, the positioning is not very coherent. For me, the

`\dot{q}`

and`\dot{x}`

are fine, but the`\dot{r}`

has its dot much too to the right and the`\dot{R}`

slightly too much to the left.I don't know how John did it with Cambria.

18 Jun 2011 — 1:45pm

My contention then is that differentiating between a "text accent" (where placement is done optically by a human designer) and a "math accent" (where the location is some mathematical center) does more harm than good. The Good, the Bad and the Pedantic.

hhp

18 Jun 2011 — 2:14pm

I just looked at U+1D445 (MATHEMATICAL ITALIC CAPITAL R) in Cambria Math and TopAccent is clearly quite to the right of the middle of the bounding box. There would be no need of TopAccent if it were always in the middle.

18 Jun 2011 — 2:25pm

> in Cambria Math and TopAccent is clearly quite

> to the right of the middle of the bounding box.

Because it was done by a human; as you said

yourself automatic positioning is incoherent.

The thing is: what's the point? What does it do

besides making it look imbalanced? What are the

chances somebody will say: "Oh, the accent is to

the right, so it can't be this rarefied letter from a

writing system I'll never run into, used right in

the middle of a mathematical equation."

hhp

18 Jun 2011 — 2:52pm

All I am saying is that I see no reason why a human being would systematically align TopAccent in the math font with the anchor that is used in the italic font to position combining accents.

18 Jun 2011 — 3:43pm

Concerning TeX, had almost forgotten that there are only two horizontal parameters given to each math character, its width and its italic correction. Even if they are fixed by a human being, that may not be enough to correctly position top accents.

19 Jun 2011 — 12:56am

@Michel, in math mode TeX positions the accents horizontally based on the kern of base glyph with the tie accent (a weird knuthian way to store that value) which is what TopAccent value in MS implementation is equivalent to.

19 Jun 2011 — 5:02am

Khaled

If I typeset

`$\hat{H}$`

in Computer Modern, the H is taken from CMMI10 and the hat is taken from CMR (the mathematical accents are always upright glyphs). To look at the metrics of the CMMI10 fonts, I need only type`tftopl cmmi10.tfm`

in any shell window. If I look at the lines concerning H, I get(CHARACTER C H

(CHARWD R 0.831251)

(CHARHT R 0.683332)

(CHARIC R 0.081248)

(COMMENT

[...]

)

)

where I skip the comment. There are three values, CHARWD, the character width, CHARHT, the height, and CHARIC, called the italic correction. That is consistent with this description of the the way

`\hat{H}`

is typeset, taken from Bogusław Jackowski's slides (GUST). Here are the two glyphs at stakeand here is \hat{H}.

No other parameter can be used because there is none.

Now if you look at the information on Opentype math fonts mentioned in http://fontforge.sourceforge.net/math.html, each math character still has a height, a width and an italic correction, but there is an additional parameter for the top accent which allows more precision. Am I missing something?

Michel.

19 Jun 2011 — 10:27am

Back to the dots, and back home where I could look at a few of my books, I can only conclude that those dots look like an editor's nightmare. I have two books (in French) published in Moscow in the seventies. In one of them one gets within five consecutive lines the following dotted q.

Both books put the dots very high above.

In some american books I have, the dots are so small and so close to the letter that I find them hardly visible.

And here is from the "Winner of the 1990 Science Book Prize", published in the UK:

19 Jun 2011 — 10:44am

"Comrade, there is dot. What is problem?"

And, to be even-handed:

"Dude, there's the dots there dude. Chill."

hhp

19 Jun 2011 — 11:30am

@Michel: The comment under the second figure is wrong, "italic correction" should be "kern" (in CM fonts the skewchar is usually the tie accent), it actually makes no sense as it is since the italic correction is a fixed value for each character regardless of what is next to it (though it get only applied when switching from italic to upright). I just checked rule 12 in Appendix G of The TeXbook and it states:

If you checked the kerning on the top of the PL file you will see:

(LABEL C H)

...

(KRN O 177 R 0.055557)

the character with octal code 177 is the skewchar. In MS implementation a dedicated field is used instead of (mis)using kerning (I guess knuth just wanted to keep the TFM file compact for the obvious reason of that time).

19 Jun 2011 — 11:45am

Hrant, it isn't a question of making the accent-like mathematical operator look either attached or detached from whatever is below them, but of the specific conventions of alignment of operators and operands in mathematical typesetting. In mathematical typesetting, one is frequently dealing with both horizontal and vertical layout, and generally speaking what is true of alignment of larger, more complex expressions is true of alignment of smaller combinations within them. Hence, the centre alignment of the accent-like operators to top of whatever is below them is the same as the alignment of the top part of an expression to the lower part. This means, among other things, that the distance between the accent-like operator and what is below does not affect the horizontal offset above an italic form. So, for instance, looking at Michel's italic H example, you can see that if the accent were raised or is something else were placed above it, this would be a straight vertical shift or alignment, not a slanted move or offset.

19 Jun 2011 — 11:48am

PS. Did you get a copy of the MS

Mathematical Typesettingbook? You can order one from Tiro for $8.50 shipping and handling (note, however, that we currently have a postal strike on in Canada).19 Jun 2011 — 1:30pm

Am I missing something?

Not really. Accent positioning is refined in a similar manner to GPOS accent positioning. Accents each have a center position defined via 'Top Accent Attachment'(optically centered in most cases) and each base can also have a center position defined; these two positions are aligned on the

x-axis, and are displayed. If a base glyph does not have an entry, then accents are simply centered. For those who don't know, accents can also be mapped to alternates via the [flac] feature ('flattened' accents, analogous to cap-variants of accents) and also to horizontal, or wide variants for placement over wider glyphs (the Math handler measures the base glyph and then uses the most appropriate accent variant). Note that the Top Accent Attachment is really only anx-axis value, unlike a GPOS anchor, which has bothx- andy-coordinates.y-positioning is dynamically achieved by measuring the base glyph as well as other elements (eg. stacked accents), whether you're in a display mode or not etc.As to fonts that support what Gus wants, he's probably already figured that out, which if you are using (Windows) Word 2007/2010 would be Cambria Math (Word defaults to this when you enter a Math zone) or Maxwell Math, when its released. For outside a Math zone, one would need a font with fairly robust GPOS accent positioning, such as Times or Arial or...?

19 Jun 2011 — 4:48pm

@Khaled. Thanks. So, there is another parameter in the kern table which is s = 56. Good! Now, here are some calculations to check:

WD = 831, IC = 81

(WD + IC)/2 = 912/2 = 456

456 + s = 456 + 56 = 512

and it is indeed true that the hat of \hat{H} is centered at 512 as one can see from this output from latexit edited with InkScape.

The GUST graphic was wrong in more than one way. Now, what is the side effect if I change the kern of H with the skewchar?

Michel

20 Jun 2011 — 4:25am

@Ross: for the math mode (zone) there is also the free XITS Math and Neo Euler and the commercial Lucida Math (when it is released, the name is not final yet).

20 Jun 2011 — 4:34am

@Michel: I don't think that kern value has any use in TeX other than controlling accent placement as kerning between, say, H and the tie accent makes no sense at all (in TeX accents precede accentee so such kern would never be applied. theoretically someone might build a font where the skewchar is some other character that get kerned but it does not make much sense and I've not seen such font.

20 Jun 2011 — 4:36am

The reason operators are that high in the russian books seems to be to make sure they are all aligned. That avoids weird looking combinations like

20 Jun 2011 — 7:07am

Khaled, thanks.