Periodcentered

Nick Shinn

1. Should it be an identical glyph to period?

2. Should it be the same character width as period?

3. Should it be centred vertically on the cap height (with suitable “optical” adjustment)?

4. Where a type is “high- (or low-) waisted” in the caps, should it be aligned with the waist?

5. Where ascenders overshoot cap height, which should it be (vertically) centred on?

6. Where ascenders overshoot cap height, would it make sense to (vertically) align it with the ascender, and provide a “case” alternate aligned to the caps?

Craig Eliason

4. Where a type is “high- (or low-) waisted” in the caps, should it be aligned with the waist?

Does anybody do that with hyphens or dashes?

James Puckett

Is this glyph still in use? Or has it been made obsolete since /ldot became common in fonts?

Does anybody do that with hyphens or dashes?

It happens in fonts that have alternate all-caps punctuation.

Nick Shinn

It is used for some math functions, James.
And in dictionaries to indicate syllable breaks.
I often used it in all cap settings as a spacer, similar to the way people use the bar.

I believe that in Catalan, manuscripts are more often keyboarded with /l followed by /periodcentered, rather than using /ldot. With this in mind, I create a glyph “periodcentered.ldot” which substitutes in the local feature. AFAIK, the only drawback would be in a Catalan dictionary.

Chris Lozos

Some people use it as an optional bullet subordinate to a full dark bullet.

Sandy Cerovich

...and in line with what Chris said, I always center it vertically with the bullet.

John Hudson

I typically make it lower, aligned with the hyphen, because it gets used between lowercase letters sometimes, e.g. as a syllable separator in dictionaries. I provide a raised version for use in all caps, accessed via the {case} feature.

Following discussion with Catalan colleagues, I think it is best to provide either a {locl} alternate as Nick suggests, or ligate

l periodcentered -> ldot

The best height of the dot for Catalan is going to depend on the typeface, and the proportions of the x-height to the ascender height. Generally speaking, it is going to be somewhere around the mid-height of the l, but it should be looked at in sample words to determine what looks best for a particular design.

Nick Shinn

The ligature method becomes uneven when anything other than zero tracking is applied.
When tracking is beyond the “fail” amount, and the ligature ceases to be substituted, the spacing is even, but disproportionately large.

Kent Lew

It is used for some math functions, James.

Technically, the bullet operator for math (indicating multiplication) is a separate Unicode character (U+2219) from the centered period (U+00B7).

So, theoretically, the periodcentered can be aligned on lowercase to work as syllable separator, with a .case variant for use in all-caps, while the uni00B7 can be aligned to work with figures.

But that’s assuming precise semantic input in the encoded text stream.

[Deleted User]

The user and all related content has been deleted.

Chris Lozos

James, I'll bet you "like the smell of Napalm in the morning", too ;-)

[Deleted User]

The user and all related content has been deleted.

Chris Lozos

It wasn't soon enough for me.

Georg Duffner

Just for completeness about periodcentered in catalan: it’s always at a syllable boundary (only occuring in the combination l·l), so in a dictionary you wouldn’t need to add an extra symbol, however all my catalan dictionaries (mono- and bilingual ones) don’t show any syllable boundaries. The only comparable symbol in one of them is a vertical bar that shows boundaries between the word’s root and affixes.

To add to John’s comment: it’s also necessary that the localised periodcentered doesn’t take up much space because the distance between the two "l" in "l·l" should be similar to the one in "ll".

One further observation: a similar case exists in occitan where you might find the combinations s·h, n·h and l·h. They’d probably need a similarly localised glyph.

Nick Shinn

@Goerge: To add to John’s comment: it’s also necessary that the localised periodcentered doesn’t take up much space because the distance between the two "l" in "l·l" should be similar to the one in "ll".

Right, a feature of this method is that no kerning is required, just adjust the sidebearings until width of the combination of glyphs = the ldot width.
But you will probably need separate “periodcentered.catalan” glyphs for UC, lc and small caps.

Georg Duffner

@Nick: But you will probably need separate “periodcentered.catalan” glyphs for UC, lc and small caps.

Exactly. The glyphs for UC and sc will be of zero width and have a negative left sidebearing.

Nick Shinn

I give the l.c. Catalan dot zero sidebearings. That seems to work quite well, and makes it easy to calculate so that it implements with the same metrics as the /ldot glyph, for what it’s worth.

Andreu Balius

Sorry for being late to this discussion. I have just joined in.
I am Catalan, so I am fairly aware about that issue. Somehow, there are some linguistic misunderstandings on how l·l (in catalan, 'l geminada') should appear in the text.
I use identical design as period and I locate it optically centered according to both l·l. But position should be different in uppercase or smallcaps. So, I'd rather use periodcentered.case and periodcentered.smcp for those cases. If you are using ldot / Ldot / ldot.smcp, then you should arrange dot according to get a nice design for all combinations of 'l geminada'.
In Catalan the so called 'L geminada' used to be a single glyph in early typesetting. So I include within my font encoding a glyph called 'lgeminada' (for Uc, Lc and SC) and put it as a 'liga' feature for default substitution in the text. A 'locl' feature could be nice also.
I believe there is no standard way to deal with this. Since not many people cares about Catalan language, they do not consider this local issue.
As a design, the 'Lgeminada' should look as a double LL/ll but locating enough visual space between the l's, otherwise it will appear very bad in small sizes.
I use to give periodcentered some sidebearing value in case I decide to use it for other purposes in the text (as separation marks, for example).
If there is any question on this, I will be glad to offer some help.

Wei Huang

What size is a good periodcentered compared to the lowercase letter weight or the period? I had read from some Catalan readers that it shouldn't be the same size in certain cases but I can't find the reference again.

Wei Huang

Reviving this long dead thread, is there a difference between dot operator U+22C5 ⋅ and bullet operator U+2219 ∙ 

⋅∙

Eli Heuer

U+22C5 is used as a multiplication symbol when typesetting math, there are many usage examples here: https://tex.stackexchange.com/questions/19180/which-dot-character-to-use-in-which-context

Kent Lew

These two characters, and their relationships to similar punctuation, are discussed in Chapter 22 of the Unicode Standard — specifically Section 22.5 in the subsection on Disunifications, with the corresponding Table 22-5, pages 807–8.

Kent Lew

I just re-read something I wrote above almost five years ago, and realized I mistyped.

What I believe I would have meant was:

So, theoretically, the periodcentered can be aligned on lowercase to work as syllable separator, with a .case variant for use in all-caps, while the uni2219 can be aligned to work with figures.

The periodcentered should be encoded as U+00B7. Which makes the second half of my original statement contradictory. It’s U+2219 which is used for math and should align with figures.

Wei Huang

@Kent Lew Thanks Kent, I read the section and I'm still not sure at what height, size and what spacing is ideal for these glyphs:

Mathematical operators render on the math centerline, rather than the text baseline. 

The distinction between middle dot and dot operator deserves special consideration. dot operator is preferred for mathematical use, where it signifies multiplication. This allows for rendering consistent with other mathematical operators, with unambiguous character properties and mathematical semantics. middle dot is a legacy punctuation mark, with multiple uses, and with quite variable layout in different fonts. For the typographical convention of a raised decimal point, in contexts where simple layout is the priority and where automated parsing of decimal expressions is not required, middle dot is the preferred representation

So I gather from this that the 22c5 Dot Operator, should be designed to work between figures and letters, and that 00b7 middle dot should align with lowercase letters (?) but in practise, in the following examples, it is inconsistent. Mostly designers make 00b7 middle dot, 22c5 dot operator the same.

Menlo:



Cambria Math:


Georgia:


Apple Symbol:


Arial Unicode MS:

Kent Lew

but in practise, in the following examples, it is inconsistent. Mostly designers make 00b7 middle dot, 22c5 dot operator the same.

Yes, that’s the reality. Partly, I believe, due to the fact that users aren’t consistent with their use of 0x22C5 and 0x2219.

Or are users inconsistent because font manufacturers have often made them the same, so there’s no visual distinction? ;-)

Michel Boyer

If you want to see what the math symbols are expected to look like for publishers that heavily use math fonts, you can have a look at the STIX fonts.

Mark Simonson

In the STIX fonts, all four are positioned the same, slightly above the center of the x-height.

John Hudson

Note that the height the STIX Two characters sit when input in a typical text editor or page layout app isn't necessarily the height they sit at when used in math equations. The MATH table includes an alignment property used in equation layout, which aligns operators to the midline of the numerals.

My own practice for typical text faces is to position the /periodcentred/ U+00B7 glyph near the mid-x-height, aligned with the hyphen. I position the larger /bullet/ U+2022 higher, typically aligned with the crossbar of the uppercase H, since the bullet is most often used before a sentence initial cap.

Michel Boyer

Here is how they render with XeLaTeX with \setmathfont{STIX Two Math} and equation   \[ x\cdotp x\cdot x\bullet x\smblkcircle x\]
\smblkcircle belongs to a series of black circles of various sizes.

Michel Boyer

Here are the relative positions of periodcentered, hyphen and cdot in commonly used LaTeX packages. 

Notice that the  vertical positioning of cdot relative to periodcentered (on any text on the current line) does not seem to be affected by the surrounding equations as suggests the following test:

(I did not check the specs). It is the whole line that moved down with the equation. Here I used Fourier (and three cdots).

Claudio Piccinini

I bookmark this old discussion as I am making final decisions on how to align these various glyphs (punctuation ones and mathematical equivalents) in my own typefaces.