GuyA,

Nothing strange. SAS just sets it to a .I, that is missing I. A simple experiment confirms it:

290  data _null_ ; 291     c = divide (1, 0) ; 292     is_c_null_i = (c = .I) ; 293     put is_c_null_i= ; 294  run ; is_c_null_i=1

So, as any missing value, it is *printed* as whatever character follows the period in the missing value literal definition. Thus, if missing value is ._, the underscore is printed, if it is just .(dot-blank, standard missing value), then a space is printed, if it is .A, then A is printed, and so on to .Z, when Z is printed. And yes, any missing value is a SAS number selected by SAS from a number of real binary representations called NAN, i.e. "not a number". Double pun intended.

Kind regards ------------ Paul Dorfman Jax. FL ------------  -------------- Original message from GuyA <guya.carpenter@GMAIL.COM>: --------------

> All interesting, thanks. > > I didn't know there was a divide function. Just a quick test though to > check to myself that it exists seemed to bring up an unusual result: > > Code: > > data test; > input var1 var2; > var3=divide(var1,var2); > put var3; > datalines; > 1 2 > 2 1 > 0 0 > 0 1 > 1 0 > ; > > Output: > > 74 input var1 var2; > 75 var3=divide(var1,var2); > 76 put var3; > 77 datalines; > > 0.5 > 2 > . > 0 > I > > The last observation output is not a number. It's the "pipe" > character. And strangely, the variable was still numeric. > > Any ideas?