228 Journal of Agricultural Research voi. v, no. t 



gave results larger than the use of all verticals. Except for flumes with 

 vertical sides, however, only 7 per cent of the results were larger. In 

 vertical-sided flumes one-third of the results were larger, so that while all 

 experiments on flumes gave an average variation of 0.9 per cent, the mean 

 of all variations was —0.6 per cent. 



Table V indicates that in flumes or lined sections such as are usually 

 used for canal-rating sections, the observation of velocities in from 12 to 

 20 verticals will give an increased accuracy of about i per cent over the 

 results obtained with from 6 to 10 verticals, and about 3 per cent greater 

 than with from 3 to 5 verticals. Under the most favorable conditions 

 where the rating curve will remain fixed, the measurement of from 12 to 

 20 verticals, depending on the size of the section, may be warranted. 

 Where the rating may be affected by channel changes during the season 

 or under such conditions as are usually obtainable in the field, measure- 

 ments based on from 6 to 10 verticals should represent good practice. 

 The use of from 3 to 5 verticals will give results as closely as the rating 

 curves derived can be applied to changing channel conditions in many 

 cases and may be sufficiently close for some purposes. Using 8 verticals, 

 only one-seventh of the results differed by more than 2 per cent; and 

 using 4 verticals, only one-sixth differed by more than 5 per cent. 



In the more irregular earth sections larger variations were found. 

 This is to be expected, as in these the velocity and depths both change 

 more rapidly near the sides than in the case of flumes. The use of an 

 average of 8 verticals in earth sections gives results of similar accuracy 

 to those obtained with only one-half as many verticals in flumes and 

 lined sections. The use of an average of only 4 verticals gives results 

 with average differences of nearly 9 per cent, and the variations of single 

 experiments are much greater. It would appear that to obtain equal 

 accuracy in gagings in earth sections with those secured in flumes about 

 twice as many verticals should be observed. The number used will 

 depend on the accuracy desired and the size of the canal. Less than 

 from 6 to 8 verticals can not be recommended, and probably 8 to 12 

 would represent good practice. For more accurate work from 15 to 20 

 may be used, although where great accuracy is desired the measurements 

 should be made in regular rating sections. Using 8 verticals, only one- 

 tenth of the experiments differed by more than 5 per cent; using 4 ver- 

 ticals, one-third of the results differed by over 10 per cent. 



A comparison of these results with those given for the different methods 

 of observation of the velocity in the verticals can be made to determine 

 the relative advantages of using either more verticals or taking more 

 points in each vertical. The use of the 0.2- and 0.8-point method gave 

 results averaging 0.7 per cent too high. The use of an average of 8 

 verticals in flumes and lined sections gave an average of 0.6 per cent too 

 small. The use of 8 verticals obtained with the 0.2 and 0.8 method 

 would tend to balance these errors, and in many cases might give as 



