If the number is not a perfect square, square root always converts the result to float. b is not an integer. It is float. And there are digits after the decimal.
0j6":b 35136418287882.219000 But when b is squared only the first 16 (approx) digits are kept. 0j3":b^2 1234567890101024100000000000.000 a 1234567890101020405060708090 b goes far beyond the precision of float double. So large values for error beyond the 16 significant digits is no surprise. ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
