Thanks! If I was using a better font I might have spotted that myself.
I know what you mean. I recently noticed something I'd written some months ago, thanking someone: "Thank's" --- aaarrgh!
Though, in my case, I blame it on the ubiquity of the "thank's" misspelling corrupting me. I fear it's only a small matter of time before I mix "their" and "there!"
Being able to distinguish between characters that in many fonts look nearly or exactly the same is really important. Adding these three lines (or something similar that shows this stuff) would make the comparison much more useful:
1. Do some of the screen shots have the wrong image size in the HTML? One or two look a little distorted to me. I.e. andale with clear type is missing the top of the text "first name" as though the image has been shrunk, but when I look at the saved image by itself, it looks fine.
2. Including a sample of all alphanumeric characters might have been a good idea. You mention things like "look at the 'W'", but there's no W in the sample. I don't want to download and install just to look at a problem.
For some of them, it's hard to say whether the images are distorted, or whether cleartype is doing a bad job. For example, DPCustomMono2 has lower case "i" that looks like it's bold, Inconsolata has "m" that doesn't look much like an "m" at all, and the "l" looks to be in bold.
Doh! I missed the Ws, sorry. Still, it'd be nice to see samples of all alphanumerics.