Xref: utzoo sci.crypt:2746 comp.lang.c:26659
Path: utzoo!utgpu!jarvis.csri.toronto.edu!cs.utexas.edu!samsung!think!mintaka!mit-eddie!bu.edu!lectroid!zircon!cme
From: cme@zircon.sw.stratus.com (Carl Ellison)
Newsgroups: sci.crypt,comp.lang.c
Subject: Re: New(?) encryption algorithm
Message-ID: <849@lectroid.sw.stratus.com>
Date: 7 Mar 90 16:41:33 GMT
References: <1877@bruce.OZ> <100295@linus.UUCP>
Sender: usenet@lectroid.sw.stratus.com
Reply-To: cme@zircon.sw.stratus.com (Carl Ellison)
Followup-To: sci.crypt
Organization: Stratus Computer, Software Engineering.
Lines: 19

In article <100295@linus.UUCP> hal@gateway.mitre.org (Hal Feinstein) writes:
>Of all the subjects to get hung up on, someone in the class
>asked if in counting 2-grams (bigrams) you shoud count every pair
>of letters or to tally each pair by sliding a pointer down one
>character at a time and taking the character pointed to and the one
>to its immediate right as a pair. [...]
>
>Now, I put it to you in networld: which is the "correct" way to
>count bigrams? 
> 


The answer is:  BOTH

If the resulting statistics are not identical (by standard statistical
tests), then you've discovered that your source text is naot stationary
-- a very interesting result which should be investigated.

--Carl