I beleive these are typos in the 1.5 spec (http://www.khronos.org/files/collada_spec_1_5.pdf):
On page 32, you have INTERPOLATION[v+1] and INTERPOLATION[n+1]. Unless these are 1-based indexing rather than zero-based indexing I believe both of these should be n.
On page 33, the OUT_TANGENTS are shown in the figure as going the opposite direction. This is misleading and I believe erroneous (at least according to the equation).
pg 32: INTERPOLATION: first occurrence should be INTERPOLATION[n+1], not [v+1].
However, n+1 should be correct. Per text, interpolation [n] is the *beginning* of segment n, so for n segments, the *beginning* point of the last segment is interpolation[n]. Therefore the *end* of the last segment where it attaches to the first segment would have to be [n+1]. (so INTERPOLATION  and INTERPOLATION [n+1] start the same segment?) Otherwise, if segment 1 starts with interpolation , then the last would be [n] but the whole paragraph shd be adjusted.
pg 33: Mark? or Remi?
Pg 33: The figure doesn't label the tangents but the tangent arrow facing left from P1 should be coincident with the tangent arrow facing right from P1 in order to form that piecewise curve. A different pair of curve segments could better illustrate the tangents if they were not coincident.
On page 34, I think that T0 = 3(P1 - C1) is a typo. E.G. compare to the notation on page 35. The left-hand side should be T1.