## What is the protocol for making modifications to someone else’s proof to prove something slightly stronger?

I have a need to modify Erdős' proof of the Sylvester-Schur Theorem to prove something stronger. See my working document at http://math.rudytoody.us/ or http://math.rudytoody.us/OppermannTheorem.pdf

If I have to modify most of the proof, I will use the entire proof (with proper attribution, of course.) However, I don't believe I will need to do that. So, how much should I show of the original? Could I do a line-by-line comparison of the changes? If I only change a few variables, could I do something along the lines of, "By changing variables a, b, c and relaxing condition x, it's easy to see that Erdős' proof arrives at the same conclusion without breaking the original."

Some suggestions would be appreciated. Thanks.

-
Short answer: "It depends on who your audience is, and what they might expect of you" – Yemon Choi Aug 16 2011 at 1:06
Most people in this audience to not have the Erdös paper handy, so probably your including a complete proof would be best. – Gerald Edgar Aug 16 2011 at 1:34
Erring on the side of caution is best here: you can always write out the whole proof; if an editor or referee feels it's unnecessary, they will probably point it out. – Thierry Zell Aug 16 2011 at 2:09
I suggest you write your own proof completely, then mention that your proof is a modification of the proof of Erdös, then cite his paper. – Joel Reyes Noche Aug 16 2011 at 2:47
I feel this question ought to be made Community Wiki, as it is asking for advice, for which there probably won't be a definitive (much less mathematical) answer. Therefore one wants a list of answers, sorted by upvoting, making CW the right course of action. That would be much better than an endless string of comments. – Todd Trimble Aug 16 2011 at 12:06

I'm following Todd Trimble's suggestion and writing my comment as an answer:

I suggest you write your own proof completely, then mention that your proof is a modification of the proof of Erdős, then cite his paper.

-
Thanks, Joel. I had also suggested this be made CW. – Todd Trimble Aug 16 2011 at 12:50
I agree completely. – Igor Rivin Aug 16 2011 at 13:03
Joel: even if the question is not community wiki yet, you can make your own answer community wiki by clicking on the appropriate box. – Thierry Zell Aug 16 2011 at 13:45
Sorry, I thought that only the poster of the question could make it community wiki. I made my answer CW now. Thanks for the info. – Joel Reyes Noche Aug 16 2011 at 15:25