1943] PROBLEMS AND SOLUTIONS 123 parallelepiped, wt at = 2(a +d") and @ wtbath yt et = sa? + 0 +d) = Aa + + C8). From (1) and analogous results for the other faces, we have 24)/8b¢, (3) { cos B = (wt — at + y? — 24)/8ca, | cosy = (wt — a= 9 + 2°)/8ab. [cosa = (wt a? — y# A tentative method for finding parallelepipeds with integral sides and di- agonals is suggested by (2) and (3). Take any integers a$bSc. Take any of the representations (2) of 4(a?-+0?+¢2) as a sum of four squares with w=xzy22 =0. These numbers will correspond to a solution provided each fraction in (3) is less than 1 in absolute value and each angle in (3) is less than the sum of the other two. Examples follow: (A)a=b=2, 3, «=are cos 3, B= =90". B)a =are cos 5/9, 7=are cos 1/9. (C) a=b=4,c=1, Here 4(a?-+02+c2) = 132, which is the sum of four squares in several ways, vis. (w % 92) = (113,151), (9 5,5,1), (71,1), (86,4, 4), (7,7, 5, 3). No parallelepiped results, however, since for each of the first two representa tions at least one cosine exceeds unity, while for the third all cosines equal unity, and for each of the last two one face angle equals the sum of the other two. (The third is interesting as a limiting case where the vertices all lie on a line; the last two, where the vertices all lie in a plane.) The vertices of the parallelepiped (B), referred to rectangular axes through one vertex, are 3, w 10. 4v 0, 0, 0), 3, 0, 0), peep ened ( ) ¢ ) Bian 2/5 14 2/5 7 2), G > 2, (5, 2V3, 2), (2, 2V, 2). A Locus Determined by Two Fixed Circles E 526 [1942, 404]. Proposed by R. C. Yates, Louisiana State University Find the locus of P if the angles formed by the tangents from P to two fixed circles are equal. Solution by Howard Eves, Syracuse University It is easily shown, by means of similar triangles, that the distances from P to the two centers are proportional to the respective radii. Hence P describes the124 PROBLEMS AND SOLUTIONS (February, circle of similitude of the given circles. Also solved by F. A. Alfieri, D. H. Browne, L. M. Kelly, J. Rosenbaum, P. D. Thomas, and J. B. Welchons. Browne remarks that this problem suggests another: “An angle of given size moves so that its legs are always tangent internally (or externally) to two given circles. Find the locus of the vertex of the angle.” J. H. Butchart and A. K. Waltz interpreted the given problem in this man- ner, and found a locus consisting of two limagons, each having double contact with both the given circles. By varying the size of the angle (while keeping the circles fixed), Butchart obtains a family of such limacons; the locus of nodes is the circle of similitude of the given circles. ADVANCED PROBLEMS ‘Send all communications about Adoanced Problems and Solutions to Otto Dunkel, Wash- ington University, St. Louis, Mo, All manuscripts should be typewritten, with double spacing and with margins at least one inch wide. Problems containing results believed to be new or extensions of old results are es- pecially sought. The editorial work would be greatly facilitated, if, on sending in prob- Jems, proposers would also enclose any solutions or information that will assist the editors in checking the statements. In general, problems in well known textbooks or results found in readily accessible sources will not be proposed as problems for solution in this depart- ment. In so far as possible, however, the editors will be glad to assist members of the Asso- ciation in the solution of such problems. PROBLEMS FOR SOLUTION 4070. Proposed by P. Erdis, Princeton, N. J. Let p denote the length of the radius of the inscribed circle of the triangle ABC, let r denote the circumradius and let m denote the length of the longest altitude. Show that p-+rm. 4071. Proposed by Harry Langman, Brooklyn, N. Y. Setting 1 + 1 + 1 part oa + ga show that bo, — by bay (= 1) | —(n—1), bo, -b, (= 1), 0 (= 2), by (= 1) | 0 0 0 bo 4072. Proposed by Richard Bellman, Brooklyn College Show that