232 Journal of Agricultural Research voi. v. no. e 



In 96 measurements the 0.2- and 0.8-depth, or two-point, method gave 

 results averaging 0.73 per cent too high, and the 0.6-depth, or single- 

 point, method gave results 4.80 per cent too high. The average variation 

 for a single measurement was 1.5 per cent for the two-point method. 

 If the results for the single-point method are corrected by — 5 per cent, 

 the average variation of a single observation is 2.5 per cent. 



In 55 measurements the vertical integration method gave results 

 averaging 0.76 per cent too high, and an average variation for a single 

 observation of 2.07 per cent. The use of three-point methods gave errors 

 greater than the two-point method alone. 



There were no marked variations of the accuracy of any of these three 

 methods due to difference in velocity, depth, or value of n in Kutter's 

 formula. 



In 92 measurements to determine the coefficient to be used to reduce 

 the maximum surface velocity as measured by small floats to the mean 

 for the entire cross section, the coefficient was found to vary with the 

 value of n in Kutter's formula and the size of the canal. For water 

 cross sections of over about 35 square feet the coefficient remains con- 

 stant for any given value of n. A table is given for the coefficients for 

 the range of conditions covered by the measurements. The coefficient 

 varies from 0.60 to 0.91 for different conditions. The average variation 

 of the coefficient for a single observation from the mean values was about 

 6 per cent, and in one-fifth of the observations exceeded 10 per cent. 



In 89 experiments on the use of observ^ations of varying numbers of 

 verticals across the width of canals, it appears that in uniform cross 

 sections, such as flumes or lined canals, observations in 8 verticals give 

 an average within i per cent and in 4 verticals within 3 per cent of the 

 discharge obtained with 16 verticals. In earth canals observations in 8 

 verticals give an average within 3 per cent and 4 verticals within about 

 9 per cent. For equivalent accuracy about twice as many verticals 

 should be observed in ordinary earth sections as in uniform lined sections. 



It was found that the use of only 2 verticals located from one-fifth to 

 one-sixth of the width of the water surface from the sides of the section 

 in canals with vertical sides such as flumes, gave results within an average 

 of 2.5 per cent. In concrete-lined sections with sloping sides similar 

 results were obtained where the velocities were measured at from one- 

 fifth to one-fourth of the width from the sides, and the areas were secured 

 from the known cross sections. 



In earth canals 2 points from one-fifth to one-fourth of the width of 

 the water surface from the sides give velocities varying from the mean 

 of the whole cross section by about 6 per cent. Where the depths at 

 these two points are used to give the average depth, the total discharge 

 is determined with an average error of about 6 per cent. Errors in 

 individual experiments were much higher. 



