I'm sure Ray Vickson is much better at this than I, so I hesitate to query his point, but.
I come at this sort of problem from a user's point of view rather than a mathematician's. If I look at an example such as this from WikiP's article on SD:
"As a slightly more complicated real-life example, the
average height for adult men in the
United States is about 70 inches, with a standard deviation of around 3 inches. This means that most men (about 68%, assuming a
normal distribution) have a height within 3 inches of the mean (67–73 inches) – one standard deviation – ..."
and look at what it tells me, I do not find any need to convert to variance. It tells me the average height is 70" and 68% of men have height within 3" of that.
Now if I want to know what that is in metres or even just in feet, I simple have to convert the units of the statistics. 70" is about 5.833 ft and 3" is 0.25 ft. ANY other values - such as obtained by squaring, dividing by 12 and square rooting again (whch I guess is not exactly what RV is saying?) - must be wrong, since the statistics are telling me about a physical fact. It may well be that RV is telling you to square the SD then convert by the square of (Edit: the conversion factor) then square root back again, but that is a pointless exercise, since it will not give you a different answer.
IMO your problem is in the conversion itself. You are confusing the conversion of Centigrade degrees to Fahrenheit degrees, with the conversion of values from the Centigrade Scale to the Fahrenheit Scale. The mean is a value on the scale, but the SD is a difference measured between values on a scale. Since it is HW, I'll leave it at that for now, but (Edit: I'll) keep watching.