As Barkley says, we have to look at what Say said and what Say did not
say.  At the end of the day, Say's Law, in its Classical (not
neoclassical) form looks like the national income accounting identity
(more or less). Production generates income sufficient to purchase that
output. 'Sufficient to buy', but that is not the same thing as 'will
necessarily be bought'. In the neoclassical version, interest rate
variations guarantee that any income not spent on consumption will be
invested, but I have never seen anything like that in Ricardo or Smith.
The Classics believed savings would generally be invested for various
reasons, some institutional (social class), others social-psychological
(to better one's condition so to be the object of sympathy), but they
had no *theory* that guaranteed that all income generated by productive
activity would necessarily be spent to purchase the entirety of that
output. Furthermore, and most importantly perhaps, there is nothing in
the Classical version that says that this is operating at the full
employment level of output.  So production might generate income
sufficient to purchase the entirety of that output, but the level of
output is not full employment. All the Classical authors recognized the
phenomenon of unemployment.  In the neoclassical version, not only do
you have Say's Law in the strong form of 'will necessarily be bought'
but there is also the tendency to full utilization of resources,
including labor (under certain assumptions, etc.).  

The more I think about it, the more the real economy seems better
depicted in the Classical authors, rather than the neoclassical.  And
despite their own stated distance from what they termed the 'classics',
in many ways Keynes, Veblen, Schumpeter, and others seem closer to the
Classics.  None of them were 'right' about everything; in fact they were
probably wrong more than they were right. But their mistakes are still
more helpful to me in understanding the contemporary political economy
than Freakonomics or whatever is in fashion these days.

Mathew Forstater