Erik Kwakkel sends a tweet linking to this Leiden blog post, about a sheet they found which shows evidence of having been used centuries later as a frisket to print characters in red. You can see the little windows which have been cut into the sheet which would be interposed between the paper (already printed with black) and the type. The full width of the type printed in red can be seen overprinting the original manuscript text and illustration on the frisket sheet. Where the little windows were cut the red ink would get through and print on the sheet being pulled. For the first pass through the press (or second if the red was printed first which actually seems more likely) all this type would have been inked in black. I’m not sure how, when printing the black they’d prevent its printing where the red was ultimately to go. There’s no way to cut a frisket with holes all over except in the few spots where red was required. Maybe they’d glue little patches over these characters after inking — though I’ve no idea how they’d prevent such slips falling off or moving and they’d have to do that after every inking which sounds ludicrously labor-intensive, even in times of cheap labor. Setting up two versions of type would be prohibitively expensive at a time when type was cast by hand and a printer’s holdings would be kept to a minimum. Really I think they’d have had to print the red first, using this kind of frisket, and then remove these bits of “red” type and replace them by quad spaces;* that way you could avoid having to reinsert these characters afterwards for the red printing.

This example, from a School of Advanced Study, University of London study of early modern frisket sheets, looks like the red ink was applied as a solid block, which would be hard to imagine unless it were being used in the inking phase rather than when the impression was pulled. Probably it just looks like this because so many impressions were run that slight variations in registration built up to fill in all the gaps between the type.

_________________

* Quad spaces are less  tall than the type, so that when ink is rolled across the type none of it adheres to the quads, leaving the area they occupy blank on the printed sheet. You can see them rather well in this picture from Paper Wren Press.