Originally by Jim Propp; see the main SRAT page. Adapted as an HTML form by Eddy: each question gets a mutually-exclusive family of radio buttons, one per answer; each answer gets a check-box alongside its radio button. Without JavaSpit, the check-boxes can serve as some space to note which answers can't be right, which is all the help you should need in working out the right answer. (If your browser supports the CSS3 :checked pseudo-class, selecting a check-box should hide its radio button.)
If you have JavaSpit enabled: selecting a checkbox will disable (and
uncheck) the accompanying radio button; and there will be some check-boxes at
page's end to turn on some extra assistance. One tells you how many of each
answer you have given or could give (to save you assorted counting); the other
tells you whether your existing answers contradict or agree with your selection
of a radio button to indicate an answer to a question – pink says your
answer contradicts some other, white says your answer agrees with the rest of
the form and greys say there's some doubt. Note that the test result is only
valid at the time it's done; no attempt is made to keep it up to date as you
change answers to other questions (aside from #20). Since
the checking script merely encodes the questions, it has
no knowledge on #20, so selecting any answer to #20 provokes a check across all
questions and uses its result to decide #20's. You can, thus, get an all
white
answer without actually being right (see Jim's comments on #20): in
fact, there are four consistent sets of answers. They only differ in #14 and
the last three questions, so it is fairly easy to work the others out, given any
one of them.
Jim said:
The solution to the following puzzle is unique; in some cases the knowledge that the solution is unique may actually give you a short-cut to finding the answer to a particular question, but it's possible to find the unique solution even without making use of the fact that the solution is unique. (Thanks to Andy Latto for bringing this subtlety to my attention.)
I should mention that if you don't agree with me about the answer to #20, you will get a different solution to the puzzle than the one I had in mind. But I should also mention that if you don't agree with me about the answer to #20, you are just plain wrong. :-)
You may now begin work.

Originally
by Jim Propp; see
the main SRAT
page. Adapted as an HTML form by Eddy with help
from Hixie.