ABSTRACT
The number of normal equations which is to be solved in the first solution by the method of correlates is equal to the number of redundant observations. In the second solution by the method of parameters it is equal to the number of necessary observations. Conversely, when the redundant observations outnumber the necessary observations the conditions should be expressed as explicit equations of the observations by means of parameters. Obviously the method of parameters is then to be preferred to the method of correlates. One should, however, be on the alert for rank deficiency problems. The solution with the correlates may be less convenient in many cases in practice, theoretically it is indispensable for a proper insight into the application of the method of parameters, especially when rank deficiencies occur, and the matrix of the normal equations becomes singular.
