Errata for "The William Lowell Putnam Competition 1985--2000: Problems, Solutions, and Commentary" by Kedlaya, Poonen, Vakil [From Richard Stanley, received 5 Nov 03] page 68: the "related question" is incorrect: consider the matrix 1 1 0 0 More precisely, the condition that each traversal sums to m should be that each row and each column sums to m. Then this statement indeed becomes the discrete version of Birkhoff-von Neumann. [From Frederic Latour, received 9 Jan 04] page 91: in Figure 8, the fraction 3/2 at far right should be 2/3. [KSK, posted 28 Jun 04, updated 25 Jul 09] page 279: The second "related question", asking whether the set S = {a^2 + b^2: a, b in Z} contains an arithmetic progression of length k for any integer k >= 1, has been resolved. Ben Green and Terry Tao recently proved that any subset of the prime numbers of positive relative upper density contains arithmetic progressions of arbitrary (finite) length. Since the set of prime numbers congruent to 1 mod 4 has relative density 1/2 within the prime numbers, and any such prime can be written as a^2 + b^2 for some a and b, this implies that indeed S contains an arithmetic progression of any finite length. See B. Green and T. Tao, The primes contain arbitrarily long arithmetic progressions, Annals of Math. 167 (2008), 481--547. [KSK, posted 30 Jun 04] page 126: the last two lines of the statement of B4 should be in boldface. [BP (from his mother), posted 8 Nov 04] page 281: the label P_8 is missing from Figure 42. [From Joe Keller (via RV), posted 21 Nov 04] page 285: the "stationary phase approximation" is due to Kelvin. [From Robert Nakamura, posted 25 Jul 09] page 229: problem 1996B5 has recently been generalized to k-balanced strings, in which every substring T of S satisfies -k <= Delta(T) <= k. See E. Georgiadis, D. Callan, and Q.-H. Hou, Circular digraph walks, k-balanced strings, lattice paths and Chebychev polynomials, Electronic Journal of Combinatorics 15 (2008), article R108.