Or.....
Add 40. Double it. Knock off a tenth and subtract the 40 you added. Very easy to do in your head.
ie. 20+40=60. Doubled = 120. Knock off 12 (one tenth) = 108. Take away that 40 you added and you're left with 68.
It also works backwards, as add 40. Divide by 1.8 then take away your added 40.
Note: double it and take away 10% is the same result as saying multiply by 1.8

@Petter Or divide by five and multiply by nine, then add 32.

@Petter, @Fernapple Us Americans are way too lazy for all that. Let everybody else figure it out. JK I can multiply 1.8 in my head.

@Fernapple. To divide by 5, then multiply by nine is tricky. Try it with 17 degrees.
It's much easier to double it and knock off 10%.
eg. 17 doubled is 34. Knock off 10% (3.4) and you easily get 30.6.
Furthermore ....
By adding 40 at the beginning, and taking it off again at the end, you do not need to add (or subtract) the 32.
The logic here is superbly elegant. It is based on the fact that at -40 degrees the two scales are in agreement.

@Petter Nah. 17 divided by five is 3.4 times 9 equals 30.6 while adding 32 is no problem at all 62.6. Just did that in my head no problem at all. As with all maths there is a load of ways, you just pick one you like.

@Fernapple now try it with 76.32 degrees. In your head, mind.

@Petter 169.376 Yes I needed a bit of paper for that, but in pratice I never really need temps to fractions of degrees, and if I did it would be in a technical working situation where i would have pen and paper to hand at least. But that one is far from easy using your method, anyway, and doing five calculations, be they easier ones, when I only need three, seems an extra burden most of the time. Moreover the conventional method promotes understanding of the problem, and is easier therefore to check for errors and reverse.

@Fernapple Actually, from a coding point of view, the add 40, select multiply or divide by 1.8, subtract 40 is a beautifully simple code.
You method has to decide whether to add or subtract 32 at the beginning or the end, then select the multiply or divide by 1.8 requires basically two sets of code.
The add 40 also gives people a real understanding of maths concepts, especially the value of zero!

@Petter Yes but your method does not work in reverse, going from F to C, if just apllied mindlessly, you still have to make choices therefore, and it is much less clear why you do so.

Besides which multiply by 1.8 as opposed to double, divide by ten, deduct result, involves a multiplication by a multiple of nine, which is what you were mainly presumable trying to avoid in the first place.

@Fernapple F-32/1.8=C

@barjoe I know, ammended see above.

@Fernapple So C x 1.8 +32=F does work in reverse F-32/1.8=C. USA never uses centigrade. Most Americans can't convert metrics in their head. I always used to say just remember of its 32°F outside it's cold. If it's 32°C outside it's hot.

@barjoe The add forty and double etc. does work in reverse if you reverse it, but my point was that you have to think about and reverse it just the same as you do with the divide by five, multiply by nine and add thirty two method.

@Fernapple In coding it's very simple. Here it is in Basic.
Input Temp ; Input (c or f) Scale
Temp =Temp + 40
If Scale = "c"; Temp =Temp x 1.8 else Temp = Temp /1.8
Print Temp - 40;
If Scale = c Print "f", else Print "c"

It's even more concise in languages like C.

@Petter As far as I know, a computer has no interest in how, easy, difficult, or silly the task it is asked to do is. It will go happily from New York to Boston via Australia if asked.

@Fernapple but the programmer, if good, does care.