Updates: Status: Invalid Comment #2 on issue 2350 by npapoyl...@gmail.com: Duration parsing nanos http://code.google.com/p/pharo/issues/detail?id=2350 It doesn't seem to be a bug since in hh:mm:ss.[0-9]+ the part after the second actually denotes the fractional part of the second and NOT the number of nanoseconds. The code actually has a nanoseconds variable which is rather misleading but it actually does what it is supposed to do: From Time>>print24:showSeconds:on: [(s + (self nanoSecond / NanosInSecond) asFloat) printOn: aStream]]. Now for '4:02:55.520 pm' self nanoSecond returns 520000000, which does the right thing in the above computation. The only reason it is used that way is because in smalltalk the fractional part of a second has accuracy up to nanoseconds so: (Time readFrom: '4:02:55.99999999999999999999 pm' readStream) printString does correctly give: '4:02:55.999999999 pm' (fractional accuracy of nanoseconds) so do the examples above since 0.520 == 0.52