View Single Post
  #5   Report Post  
Posted to microsoft.public.excel.misc
Jerry W. Lewis Jerry W. Lewis is offline
external usenet poster
 
Posts: 837
Default Excel Addition Error

I have never seen an instance where Excel's basic arithmetic was not correct
per the IEEE 754 standard. Moreover I have always found integer arithmetic
exactly match naive expectations. I have been unable to reproduce your
results despite attempts in 4 Office versions (2000, 2002, 2003, & 2007) on 3
PCs.

The most common problem when people think they see an Excel math error is
that their inputs are calculated results that are formatted to hide decimal
places that do enter into the calculation in question. Did you follow your
own directions (copy/paste from your newsgroup post into Excel and sum those
values)?

The next most common problem, is that Tools|Options|Calculation is set to
manual recalculation, so that calculations are not updated when inputs are
edited.

Another problem that can occur with multivariate functions (such as SUM())
is that text that only looks like numbers will be ignored, but I see no
opportunity in your example for that to be an issue.

If you still get your reported result after ruling out the preceding causes,
is it reproducible in other packages? My guess is that Excel's basic
arithmetic is performed in the processor chip instead of in software.
Similarly, I doubt that other packages would re-invent the wheel, so I would
expect a processor math bug to be manifested in other calculational software.
If it is reproducible in other packages, you should contact the manufacturer
of your computer to report a defective chip.

If the problem is not attributable to any of these sources, then you should
post back with the exact cells/formulas involved as well as the exact version
of Excel, the operating system, and hardware information.

Jerry

"dwright" wrote:

Hey, I'm not kidding. I'm using office 2003, and I double, triple, and
quadruple-checked everything. I initially assumed that Excel was correct,
and the error was mine, or SQL Server's. I have actual witnesses who not
only saw me do the test, but did it themselves and came up with the same
result. I wound up doing the calculation three times by hand to make sure I
wasn't seeing things.

If it is not replicating, then perhaps it is processor-dependent, or it has
been fixed in a recent patch, if so, then that's fine. I can send the
offending spreadsheet if you like.