[R] testing for differences in sums - how to?

[James Howey]

If I want to test for a salesman effect on average sale, I model it as "Sale ~ 
Salesman". But what if I want to test for a salesman effect on total sales? 
How do I model that?

Given a significant salesman effect, how do I follow up with the equivalent of 
Tukey's HSD to determine who the (indistinguishly) top salesmen are?

If I add another factor, say "region" to the design, how do I then model the 
result?

Thanks,

jkh