Measuring a typeface by hand?

james_dalgarno's picture


I'm investigating the state of modern newspaper design and the overwhelming effect screen based devices are inflicting on the newspaper industry In the developed world. I'm analysing a few papers from around the globe, and i'm looking rather intensely at the content on each page and how many words in each column etc. But I am also trying to get a rough idea of what point sizes certain papers or pages are set in. I wasn't sure if the point size of a face was the width and the height of a typeface ?? e.g Exchange for The Wall Street Journal, Measured the height @ 6 pt and found on average 4 letters per Pica, and just added them together and divided like a ratio, to get 8 pt?? Is this how it is done or am I over complicating things?? lol

I would be very grateful for any feedback?

Many thanks

James Dalgarno

Don McCahill's picture

Nope, it isn't that easy. Type is a measurement of height, but you can't just measure characters. The actual dimension is the height of the piece of metal that the type sat on, back in the days when it sat on a piece of metal.

In a newspaper, the best way is to measure from the bottom of one letter to the bottom of the letter in the line below (assuming no descenders). But even that can fail, because sometimes additional leading (or space between lines) is added. Very few newspapers used leading, however.

David Sudweeks's picture

There are a few things you can empirically measure, and I suggest you start with these: x-height, cap height, leading by which I mean the distance from one baseline to the next, and line length in characters or words (on average). Keep in mind though that type size is set by a number of factors, and at output is commonly not in whole point units. It would not be uncommon to see a newspaper's body copy set at 8.9 pt with 11.36 pt leading. (Units pertaining to the baseline grid are {more often than type size} easily divisible whole point units, such as 4, 6, or 8 pt, though you can't count on this predictably to be the case.)

Michael Hernan's picture

When comparing types for a similar function I settled on equalising the x heights of the fonts I was looking at. The size of the type is in itself arbitrary - it is how the forms make use of the space is what counts.

Q. Why the x-height?
A. because this is where the type is most complex. Capitals are much less complex, are allot more geometric and have an easy-life compared with the lowercase in the same font because they have more vertical space in which to accomplish their task.

With the x-heights equalised you can go about making direct comparisons.

William Berkson's picture

There is a certain amount of space built into most fonts. That is, if you take the "em" as the height from one line to the next, with no leading, then the distance from the highest ascender to the lowest descender will be less than the em. Because of this bit of built-in vertical spacing, there is no way from an example to be totally sure about the point size from a printed sample.

The actual distance from one baseline to the next will be the point size of the type + the leading, but how it is distributed between those two you can't be precise about without having the type in an application.

And yes, x-height is a very important design variable, though there are lots of others.

BeauW's picture

If you identify the font being used (and have the same font) you can print your best guess and see if they match. Adjust accordingly.
This would be easiest to do in Adobe Illustrator- measure any one letter, make a box in illustrator the same height, type the letter then drag till it matches the height exactly- look at the point size.

I mean, this is messy, but it should be pretty accurate.

Nick Shinn's picture

Your initial concept of "characters per area" is a good benchmark, far better than anything to do with type size, which is meaningless from a statistical point of view.

Having said that, whether the character count is for body text or decks/infographics/callouts &c. is relevant.

So perhaps a statistic on the percentage (area) of type that is text, headlines, or "other" in a particular paper is necessary.

But then you would need to factor in a White Space coefficient....

sampler's picture

I find it pretty accurate to measure from the top of the tallest ascender to the bottom of the lowest descender.. For most typefaces using the "f" and the "g" is pretty accurate.

A quick way you can measure these within an application is to type an "fg" convert to paths and read the vertical measurement of the pair.

Of course built in leading is often part of the measurement (thus 12 pt times new roman is actually only 11.8pt measuring the "fg"

russellm's picture

Try asking the newspapers about their type.

quadibloc's picture

If you identify the font being used (and have the same font) you can print your best guess and see if they match. Adjust accordingly.
I have found, though, that at least on Windows systems, with some exceptions, almost all fonts have a considerable amount of built-in leading, much more than was the case with metal type.
Basically, this is so that results, without specifying leading, for single-spaced text in a printers' typeface will, instead of looking like type set without any leading, look like single-spaced typewriting.
Maybe this isn't a property of the font itself, but of word processors, however I am aware of some fonts that provide a true unleaded appearance in the same word processors that do this with most fonts.
So playing with a Times Roman font on your computer to match the point size of Times Roman text printed with metal type is going to give erroneous results.

William Berkson's picture

While there are different specs. for vertical metrics for Windows and Mac in the font, what makes the final determination of the line gap is the software, reading those specs. For example, I believe that in MS Word they normally add 20% in leading beyond the point size. I think this is true whether Word is operating on a Mac or Windows. And in InDesign, it will set exactly the leading you specify, whether on Windows or Mac.

The specification of vertical metrics and their relations to different apps is very messy, but I believe it comes down to that for these most widely used programs.

Syndicate content Syndicate content