RE: Modulus 10 formula (REPOST)

Bob,

I'm just finishing up a credit card project, and I found that there seem to be 
as many versions of the
Mod10 as there are merchants/card issuers.  I was working primarily with 
validating merchant numbers.
One bank used only the last 6 digits of the number, beginning with the left and 
working to the right.
Another used the entire number, but if the first 3 digits fell within a certain 
range, you changed the
first digit of the number to 0, then worked your Mod10 from right to left.  
There was no way I could
use the canned IBM Mod10 function for display files.  I had to write my own 
Mod10 validation routine
for every issuing bank.  I'm not sure this answers whether there are 2 
"official" versions, but I
found that to be irrelevant, since no one used it without significant 
modifications.  I guess everyone
knows how to break Mod 10 these days, so banks are getting creative in how they 
apply it.

Regan L. Barr
Cintas - Corp. Systems
Accounting & Finance
Systems Analyst

------------------------------

date: Wed, 25 Feb 2004 08:43:10 -0600
from: "Bob Cozzi" <cozzi@xxxxxxxxx>
subject: RE: Modulus 10 formula (REPOST)

Gosh, is it too early for me to type or what??? :)

-Bob

I've been going back and forth on a Modulus10 routine because, as it turns out 
there seems to be two
different formulas: IBM and LUHN. The biggest issue is a subtle one which 
follows. With IBM, you
always start with the left most digit (highest order digit) and select the even 
position digits by
moving to the right and multiplying hem by two. This seems to be done 
regardless of the length of the
original number; therefore, 34567 would use 3, 5 and 7, and 4567 would use 4 
and 6. But the LUHN
Modulus 10 formula begins with the rightmost digit (lowest order digit) and 
selects every other digit
moving to the left, and multiples them by two.

So with IBM we would get this (starting on the left):

Start with: 3456
Step 1: Multiple 3 and 5 by 2 = 6 10 
Step 2: Add them: 6 + 4 + 1 + 0 + 6 = 17
Step 3: 20 - 17 = 3
Check Digit = 3
Result: 3456-3

But with the LUHN formula we would get this:

Start with 3456
Step 1: Multiple 4 and 6 by 2 = 12 and 8
Step 2: Add them: 3 + 8 + 5 + 1 + 2 = 19
Step 3: 20 - 19 = 1
Check Digit = 1
Result: 3456-1

So is the IBM one screwy or are there two of them?
-Bob Cozzi