Question reguarding custom rounding practice...

My supervisor has this method of rounding decimal places that is contrary to the widely accepted mathematical process.

For two place decimals the second place is always rounded down to an even number, with the exception of those decimals whose second place is '5' (0.25, 0.75).

For example:

0.1250 = 0.12

0.6250 = 0.62

0.6875 = 0.68

In the past, those dimensions that AutoCAD rounds normally, I have "fixed" by changing the Dim Roundoff Value to 0.02. This works until the dimensioned entity gets changed, then I have to remember to reset the Dim Roundoff Value to zero, which obviously creates problems if I forget to change it.

My question is; Is it possible to automate this procedure by either changing the rounding rule, or by writing an LISP that affects only the applicable dimensions?

Also, does anyone know if this practice originated from any specific discipline?

Thanks!

My supervisor has this method of rounding decimal places that is contrary to the widely accepted mathematical process.

 

Also, does anyone know if this practice originated from any specific discipline?

!

I have met with this method of rounding in surveying. When averaging lots of figures, if you always round up, you are introducing an upwards bias to the set of figures. If you always round to the even figure, there is less shift to the whole.

I expect someone else might have a more elegant explanation.

I have met with this method of rounding in surveying. When averaging lots of figures, if you always round up, you are introducing an upwards bias to the set of figures. If you always round to the even figure, there is less shift to the whole.

 

I expect someone else might have a more elegant explanation.

That makes a lot of sense. My supervisor was degreed in civil engineering and has in the past brought those kinds of methods over to our mechanical drawings.