In l941 the Tennessee Valley Authority in Chattanooga hired me as one of a group of mathematicians to convert logarithmic surveying forms to calculating machine surveying forms. It was the heyday of the calculating machines—the Marchant, the Frieden, and the Monroe. These machines were about the size and heft of the then standard typewriters. TVA chose to use Monroe machines.

The forms had to be arranged so that once the given information was inserted, any person familiar with a calculating machine could step by step arrive at the desired final result. That is, the forms had to be self-explanatory. They were also to contain a built-in check of the calculation work. The people doing the calculating didn't have to know anything about surveying. It was felt that if for a particular problem three forms should be passed out in the calculating office, and if on return all three self-checked and agreed with one another, then the final result could be unhesitatingly accepted.

Consider, for example, the important three-point problem. At that time all states of the country, and many important foreign countries, were on coordinate systems. That is, each point of a concerned region bore coordinates with respect to some appropriate frame of reference. Suppose there are three points A, B, C with known coordinates. A transit is set up at a point P of unknown coordinates and angles APC and BPC are read. From this information, along with trigonometric tables, one is to find the coordinates of the point P.