There is no single best way to design large motors or to run performance tests. Standard test codes do not yet take account of some basic test evaluation problems. Written around simple, manual methods of analyzing results, they have not incorporated statistical methods of deriving performance parameters from a few points of test data. Instrument accuracy, test condition tolerance, and manufacturing variations all combine to generate a test accuracy tolerance which cannot be precisely defined. Different manufacturers may use different standards, and it is particularly important that buyer and seller agree in advance on the type of test to be used to verify performance.