Line spacing, or leading, is a more subtle question than is sometimes acknowledged. It relates not just to the absolute height of the characters (the bare minimum usually ensures that an 'h' and a 'y' which have an ascender and descender, respectively will not touch) but also to the width of the line, and indeed to the internal shape of the characters and the overall color of the page.
Unless the font involved has exceptionally long ascenders and descenders, more than any reputable text face does, merely not-touching is insufficient leading for optimum legibility. Hence the typesetter's minimum is generally 10/12 or 12/14 a couple of points of leading. Many of us prefer more (10/13, 12/15 or 12/16) but we usually lose the fight for the extra paper. Three fewer lines a page in a 300-page book can mean another 2025 pages, and the printing cost goes up.
Web text is usually set pretty tight: The Trebuchet and Arial samples shown on a previous page are essentially 12/14, the Verdana slightly more. One of the reasons I like Stone Sans is that it shows up naturally at 12/16; one of the reasons I hate the Apple Times and Helvetica is that they appear at 12/12.
Some of the variables, especially the subtle ones of color and the overall impression of a page, are hard to judge in advance at the best of times, and most of them are impossible to control on the web. In normal practice, typesetters and indeed designers fall back on well-grounded rules of thumb, the most relevant of which to the average web user is that wide lines require more leading. If they are too wide, of course, double columns are preferred; the adjustment comes in that in-between area.
Somewhere around four-and-a-half inches (27 picas in print terminology; 324 pixels at 72 dpi) is generally considered the optimum for book text, not much more and no less than about three. Sure, this is a narrow range, but then people's eyes fall into a narrow range. For a 2835-pica column, which practical considerations may force you into, you add a little lead not a huge amount, perhaps 520%. You are not making a statement here, you are trying to help the reader.
Basically, the wider the line, the more likely it is that a reader's eye will jump to the wrong new line unless the lines are well distinguished; and so the more leading the text needs.
Web text, which is usually about 20% bigger in size than standard book text (12-pt rather than 10-pt), might naturally run a little wider, say up to five-and-a-half inches (396 pixels at 72 dpi). Beyond that, you would like to increase the leading, but you can't.
Since, on the web, we cannot at present easily specify leading, we had better specify line width. And, in practice, we had better keep it short. These pages have a text column of 350 pixels, if everything goes right between me and you; perhaps a little narrow for 12/16, but as much as I'd like to see for 12/12 (not that I'd like to see that at all).
And of course, if you run into wide text, either because the page designer has made no effort to control it, or because they have asked for, say, 67% of the screen and you have a huge monitor jump in there and fix it. Your comfort is the main thing; all these guidelines are just that.
The Electronic Telegraph site, which I like, rather cleverly keeps its left column (listing sections) constant without using frames while its right column (the first paragraph of, or the full, story) varies to screen width. Evidently it is designed, sensibly, for a 640-width screen. It's awkward at my full 800-pixel screen, because then I see about a 9-inch text width, and it would surely be weird at 1024; but if I narrow the window so it fits the fixed top logo that is so conveniently provided, the text width becomes seven inches or 42 picas, about500 pixels. This is still more than I prefer, but at 12/16 it's workable, and you do get a lot of words on the screen.
By the way, screens do vary in resolution one study of a single department in Manchester found dot pitches running from 53 to 111 dpi but what the heck, you have to assume something. Your objective reality may vary; your subjective one surely will.
There is a valid reason, I think, for double-spacing between paragraphs
on screen, and not on high-resolution printed material. At least I have
a theory about it, and offer this discussion as a simple experiment.