The proof of the next lemma depends on the following assumption, which is explicitly stated by Archimedes [2, page 3]. This assumption involves the ideas of curve with given endpoints and length of curve (which I will leave undiscussed).
Proof:
Case 1:  Suppose 
.
Case 2:  Suppose 
.  Then
Proof:  By (9.25) we have
The proof that 
 is
similar. 
The proof of the next lemma involves another new assumption.
.  
Let the tangent to the unit circle at
Remark: Archimedes makes a general assumption about curves that are concave in the same direction [2, pages 2-4] which allows him to prove our assumption.
Proof:  
Suppose 
.  Draw the tangents to the unit circle at
 and
 and let the point at which they intersect the 
-axis be 
.  (By
symmetry
both tangents intersect the 
-axis at the same point.)  Let 
 be the
point where
the segment 
 intersects the 
-axis, and let 
. 
Triangles 
 and 
 are similar since they are right triangles with a
common acute
angle.
, so by
assumption 9.35,
Proof: If 
, then it follows from 
lemma(9.36) that 
. Since
Let 
 be a sequence for which 
 for all
 and 
.  Then
we can find a number 
 such that for all 
.  
By  (9.39)
. 
, it follows from
(9.38) that
.  Here I have used the fact that 
.  Also 
 so
 for arbitrary 
I wrote a set of Maple 
routines to do 
the calculations above.  The procedure  sinsq(n)
calculates 
 and the procedure 
mypi(m)
calculates 
.  The `` fi" (which
is
`` if" spelled backwards) is Maple's way of ending an `` if" statement.
`` Digits 
" indicates that all
calculations are 
 
done to 
 decimal digits accuracy.  The command `` evalf(Pi)" requests
the
decimal approximation to 
 to be printed.
> sinsq := > n-> if n=1 then 1; > else .5*(1-sqrt(1 - sinsq(n-1))); > fi;sinsq := proc(n) options operator,arrow; if n = 1 then 1 else .5 -.5*sqrt(1-sinsq(n-1)) fi end
> mypi := m -> 2^m*sqrt(sinsq(m));
> Digits := 20;
> mypi(4);
> mypi(8);
> mypi(12);
> mypi(16);
> mypi(20);
> mypi(24);
> mypi(28);
> mypi(32);
> mypi(36);
> mypi(40);
> evalf(Pi);
 is the area of
a regular