Alan N. Graham wrote:
> As I was reading through the numerous postings on saw sharpening, a weird
> thought crossed my mind. 
> 
> There has been some discussion on whether the proper sequence should be
> joint, set, sharpen or joint, sharpen, set. 
> 
> This started me thinking about why we joint first. Setting a saw via normal
> mechanical means is approximate at best. 

Most of the pliers sets have gauges for position and amount of set.
On the "disc anvil" type these are linked. The set is referenced
to the mean top of the teeth. That all sounds (in theory) fairly
accurate to me, and appears (IME) to be so in practice.

 > Although we try for the same set on
> every (alternate) tooth, there are bound to be small differences in the set
> of each tooth. (Or in my case, sometimes not so small differences.)If all
> the teeth had been jointed to the same level, this varied set would result
> in a slightly uneven edge to the saw.

True, but only if your assumption of uneven set is true.

> 
> Suppose that we set the teeth first, before jointing or sharpening. Now the
> jointing assures that the entire edge of the saw is level (assuming you are
> not dealing with a breasted saw). 

But the commonest sequence is: joint, shape, set, file(sharpen).

The shape operation make the teeth (as far as possible) uniform, yes
uniform set (which I assert is the case, more or less) results in
a second, different, uniform condition.

> Jointing before setting results in a tooth edge which will rise slightly
> from the outside to the inside of the saw blade when seen from the end (see
> horrible ascii art below). Setting before jointing would result in a flat
> bottom tooth.
> 
> 	/\\           __
>      \\  \\          \\ \\
>       \\  \\          \\ \\
> 
>    Joint then set    Set then joint
> 
> I haven't even attempted to calculate how this would affect the other
> characteristics of the tooth.

You're ignoring the fact that filing will remove these effects. The only time
jointing could effect the final shape of the teeth is if it's done last.

In fact, you've pointed out a reason that Pete Tarans approach of setting
teeth as the final operation has a interesting consequence
on a rip saw; the edge of the tooth near the centre line of the saw
will be slightly higher than the outside, so the saw will cut a slightly
'V' shaped  groove, instead of the flat bottomed dado normally pictured.

> 
> Now, obviously, there must be a reason that jointing is always the first
> step. However, it's not immediately obvious to me. 

Jeff answered this.

> 
> Anyone like to explain what I'm missing about this picture?

I've tried my best ;-)

     BugBear