Yes thats a good point. I guess, one advantage of the keystroke programming model it is relatively easy to use for simple use cases. If you can use the calculator you can program it.

Could do, maybe as a KZfaq short 👍

They are 'ultrasonically welded'. No easy way to open them.

@@DerekDavis213 Thanks for the reply.

Yeah for some data series the 21S does return an erroneous sdev of 0- for example 999,999, 1,000,000 & 1,000,001. It seems to me the internal logic to check for potential situations where Wilford method fails is not 100% accurate.

@@CalculatorCulture mh my first guess is because in that case, sum(xi)^2/N is exact, while in the example in the video it is not. The 41C shows similar behaviour, OUT OF RANGE (!) for 1e5+(0,1,2) and 0. For 1e5+(-1,0,1)

I was wrong. It is simply so that the SDEV calculation is carried out in extended precision (12 digits for the 41, 15 for the 42 and the other Pioneers), and when the result under the root is negative, the error is thrown. Free42 also implements the rule (amazingly) but since it doesn't use extended precision, it fails for different arguments eg. 1E17+(2,0,2)

@@wernerhuysegoms9674 thanks, that's useful to know.

My videos are each about a different modern or vintage calculator.