BAROMETRIC PRESSURES ON THE GREAT LAKES 7 



the new observation equations were based upon an entirely independent 

 group of data, elevations of water surface, from a different gage, which might 

 even be on a different lake. 



Though the general procedure was that indicated above, the arrangements 

 were such, especially when several computers were working at the same time, 

 that two or more least-square computations were in progress simultaneously, 

 relating to different gages or to different phases of the investigation. In gen- 

 eral, the investigation has been so managed that each general conclusion 

 adopted depends upon two or more least-square solutions and the corre- 

 sponding studies based upon independent sets of data, usually from different 

 gages and if feasible from different lakes. 



The least-square method of solution was adopted as the principal method 

 of attack on this problem because it was apparent that several different fac- 

 tors or influences were operating simultaneously to cause fluctuations of the 

 elevation of the water surface at a gage, no one of which could be safely 

 neglected while attempting to evaluate others. In this investigation, deal- 

 ing with the facts on the full scale of nature and under uncontrolled natural 

 conditions, it is not feasible to use the familiar laboratory strategy of elimi- 

 nating from the phenomena by control the influence of all factors except one, 

 determining the influence of that one, and then doing likewise for each of 

 the other factors in turn. Hence, it was believed that the best method of 

 attack in this case is the least-square method of computation. That method 

 is especially adapted to taking into account simultaneously several control- 

 ling factors and to determining simultaneously their separate influences. 

 The outcome has fully justified this method. 



In the broader investigation of evaporation, including the investigation 

 of wind effects and barometric effects which is here reported, 74 complete 

 least-square solutions and the corresponding studies have been made. 

 Typical solutions each contained from 100 to 600 observation equations, 

 each having for its absolute term the observed change in elevation of the 

 water surface at a gage during a day or an hour, and each containing from 2 

 to 8 unknown constants to be determined. In an extreme solution each 

 observation equation contained 40 unknowns, and there were 619 observa- 

 tion equations in the set. 



Of the 74 least-square solutions referred to in the preceding paragraph, 9 

 were directly utilized in determining the wind effects and barometric effects. 

 All of the remainder contributed more or less indirectly by gradually 

 leading the investigator toward a true understanding of these effects and 

 therefore toward a true expression of the theory. 



The statement just made, indicating the large number of least-square 

 solutions made and their complexity, shows this to have been an extensive 

 investigation. More than 22,500 man-hours of time have been spent on 

 the routine part of the computations and studies connected with the broad 

 investigation of evaporation, including the investigation of wind and baro- 

 metric effects. From 1 to 10 persons have been engaged in this routine work 



