Works: Calculation Errors in Spreadsheet Functions (107462)
The information in this article applies to:
- Microsoft Works 2.0
- Microsoft Works 2.0a
- Microsoft Works 3.0
- Microsoft Works 4.5
- Microsoft Works 4.5a
- Microsoft Works for Windows 95, version 4.0
- Microsoft Works for Windows 95, version 4.0 4.0a
This article was previously published under Q107462 SYMPTOMS
Spreadsheet formulas and functions in Microsoft Works for Windows
may return incorrect results in certain cases due to a binary rounding
error. Some decimal numbers, most notably 0.1, are repeating binary
numbers. When repeating numbers are added together, the result may
not always be accurate.
The error will usually occur in the sixteenth or seventeenth decimal
place and can be ignored in most instances.
For example, the returned value when 100.1-100-0.1 is calculated is
given in scientific notation as -5.69E-15. This number is -5.69*10^-15
= -0.00000000000000569. While this number is not zero, it is close
enough for most computational purposes.
The Windows Calculator and Microsoft Works for MS-DOS return similar
calculation errors.
RESOLUTION
This error can be corrected up to 13 significant digits (14 would not
correct the error in all cases) using the ROUND function.
The ROUND function has the syntax:
For each formula or function that is returning an incorrect result,
place the ROUND function in the same cell with the original formula or
function as the "x" parameter and a number from 0 to 13 as the
"NumberOfPlaces" parameter to indicate the number of significant digits
after the decimal desired.
The following examples demonstrate some common calculation errors that
may occur and the accompanying ROUND function that will correct the
error in each case:
Formula Result Actual ROUND()
------- ------ ------ -------
=0.145*100-29/2 -1.77636E-15 0 =ROUND(0.145*100-29/2,13)
=MOD(0.28*100,2) 3.55271E-15 0 =ROUND(MOD(0.28*100,2),13)
=100.84-100-0.84 3.44169E-15 0 =ROUND(100.84-100-0.84,13)
=100.1-100-0.1 -5.68989E-15 0 =ROUND(100.1-100-0.1,13)
These errors occur when all of the bits in a binary number cannot be
used in a calculation (computers use a binary number format to store
numbers internally).
Modification Type: | Major | Last Reviewed: | 11/15/2004 |
---|
Keywords: | kberrmsg kbprb KB107462 |
---|
|