BAROMETRIC PRESSURES ON THE GREAT LAKES 



75 



rejections and combinations should be made. Aside from the special cases 

 in solution W29 at Cleveland indicated above, the rejections and combina- 

 tions appear to be due, as a rule, to seiches, and especially to the first 

 and largest oscillation of a new seiche which has just been started by an 

 unusually vigorous impulse given by a change in wind or by a change in 

 barometric gradients. 



Aside from the four final solutions for wind effects of which the results are 

 given in table No. 13, page 63, and which fixed the adopted value of C x , 

 three other wind solutions for different gage stations were also made, 

 namely, one each at Milwaukee, Harbor Beach, and Mackinaw. These 

 three gave determinations of C x which have such large probable errors that 

 the values- were not used in fixing the adopted value of C x . These solutions 

 were not made with all the refinements of the final solutions. Nevertheless, 

 they furnish a valuable check on the correctness of the theory as to wind 

 effects which has been used in the investigation. 



In table No. 17, showing the results of these three solutions, the assigned 

 weights are on the same basis as the weights shown in table No. 13. 



TABLE No. 17. 



Note that the sum of the weights for these three solutions is less than 

 0.01 as great as the sum of the weights for the four final solutions, as shown 

 in table No. 13. 



Note that the residuals for these three solutions are not large enough to 

 prove that systematic or constant errors are present. One of the three 

 residuals is less than the corresponding probable error, and in the extreme 

 case, solution W17, the residual is 3.8 times the corresponding probable 

 error. According to the laws of probability, a residual 3.8 times the cor- 

 responding error should occur once in about 100 times. 



If these three solutions, with their proper weights, as shown above, were 

 used with the four solutions of table No. 13 to fix the final adopted value of 

 C x , that value would be +0.086, differing only 0.002 from that actually 

 adopted. 



Consider the values of S x which are shown in table No. 10, page 53, and 

 note the contrast between the values for the two lakes concerned. The maxi- 

 mum value for the Lake Erie stations is 8.32 and the minimum value is 1.72. 



6 



