I don’t know for sure, but as a former ad agency type director and former advertising type shop manager, I can say that the discrepancy probably stems from advertising agency guidelines established in the 1960s and ’70s, when film setting displaced Linotypes, Monotypes, and Ludlows (the machines, not the foundries), though it may have happened earlier. During this time period, short column measures were popular, and long em-dashes clashed with the tightly-set typography that was then popular.

Agency guidelines often stated something similar to this: “en-dashes, with a thin space on either side (note: not a word space) are to be substituted for em-dashes for all composition.” Many ITC faces have short em-dashes, too, though many do not. I believe that foundries often dictated the length of em- and en-dashes, too, and that there has never been a definitive standard. How many Times and Times New Roman versions are out there—again I don’t know; FontBank’s Times Black features an em-dash which is shorter than the en-dash, and Monotype have released at least dozens of versions—but they are all based upon Victor Lardent’s drawings, supervised by Monotype contractor Stanley Morison, for The Times, a London daily, where the face first appeared on October 3, 1932; you could look at microfilms of that paper or contact Monotype to find which em-dash length is original.

Jim Craig published Y&R’s typographic guidelines, written by Klaus Schmidt and Roy Zucca, in his book, Typography for the Graphic Designer.

I hope this helps, but I’ve probably just further muddied the waters.

In metal type do dashes of either kind extend the full width of the type body?

Yes, but again, I think the standards were loose. And there are differences between hand-set foundry types and machine-set type, cast from matrices.

I remember reading somewhere, that in continental Europe, em-dashes were often cast with hairline gaps on either side, to allow for setting them flush against the type without spacing material. Whereas North American and English foundries had them run across the entire body, so that the sorts could be used to form lines.
_____________________________________________
Personal Art and Design Portal of Ivan Gulkov
www.ivangdesign.com

I believe oprion is correct, and that explains the need for the thin space in the agency specifications. And if memory serves, foundry (hand-set) types were cast with the hairline gaps, as oprion says was the case in Europe.

Those are some greatly informative answers; grist for the mill of typographic discovery. You've both made me want to go back and look at the beginnings of type before I start fumbling around with the hows and whys of modern typography.

Thanks,

Alek

@ Cameron:
"agency guidelines"!? It's amazing to read that they existed. When I was an ad agency type director in Minneapolis in the late 1980s and early 1990s, no one gave a flying fumble about what dashes looked like except me, the type director. & when I put in 3/4 em dashes on full em width, no one noticed.

@ Craig:
In metal type, em dashes shorter than full em width did exist, but were hard to find. Good Monotype shops (such as Mackenzie & Harris in San Francisco) routinely substituted a 3/4-em dash in the matrix case for the full em dash, but cast it on the full em width. That's where I learned the practice. I carried that practice with me to the early days of PostScript fonts, when we had not yet gotten the richness and sophistication we have now. But that meant that I'd most often have to do a search-and-replace to get rid of the full-width em dash and replace it with a 3/4-em dash on full em width.

My experience with foundry metal was that most faces had full-width em dashes. That seems to contradict Cameron's statement, I know.

I cannot recall seeing en dashes less than full en width.

powers

PS: I just recalled that when I worked at M&H, the owner was a Swiss guy. maybe that accounts for the 3/4-em dashes.

Powers, I stand corrected on the foundry types, it’s been a good 25 years since I’ve seen any physical examples.