92 BULLETIN STfi, U. S. DEPARTMENT OF AGEICULTURE. 



guide as the present state of knowledge makes possible. The writer thinks all 

 who have been concerned in experimental work of this character must recognize 

 that a close approach to exact agreement among results is not to be expected. The 

 author's analysis of the conditions affecting the reliability of the experiments is very 

 thorough and his work seems to merit acceptance equally with the best previous 

 results. Nothing occurs to the writer by way of criticism or discussion worthy of 

 publication. 



Author's closure. — The apparent lack of conformity in results of hydraulic 

 experiments has discouraged many observers. It would appear that carefully con- 

 ducted experiments, made with the best of apparatus, should give results of extreme 

 consistency. It is more than likely that results that should be consistent are so and 

 that variations are due to influences that can not be seen or guarded against. The 

 best that may be done is to anticipate all known abnormal conditions and either elimi- 

 nate them or make corrections for them. 



Discussions of some experiments mentioned appreciated field difficulties. There 

 is a vast difference between testing small pipes, weii's, or other devices in a hydraulic 

 laboratory, where all of the factors are assuredly in the control of the observer, and 

 testing large commercial structures in the field under var jdng weather conditions with 

 unavoidable fluctuation in discharge of water and other difficulties that may or may 

 not come to the knowledge of the experimenter during the tests. Seldom it is that a 

 field layout approaches ideal conditions for experimentation. Except in rare cases 

 tests of commercial plants must be foregone if laboratory conditions are to be met. 

 When apparatus is set up for test in a laboratory, minute measurements of diameters, 

 water volumes, and other factors may be made. Both interior and exterior may be 

 examined ydih microscopic care. In field tests of pipes, on the other hand, it is not 

 often that water may be withdrawn for sufficient time to enable the experimenter to 

 fully acquaint himself with conditions that are ordinarily hidden. 



Probably the most accurate series of tests of record are those conducted by Saph and 

 Schoder on brass pipes. The largest of these was slightly more than 2 inches in diam- 

 eter. With all the accuracy the experimenters could bring to bear upon these tests in 

 a hydraulic laboratory, the results were such that they qualified their derived formula 

 with a factor of ±7 per cent.^ 



When the above qualification is necessary for a formula based on laboratory tests on 

 pipes having uniform structural characteristics which have never been subjected to 

 deposits, growths, or tuberculations, what is to be expected of a formula for use in the 

 design of wood-stave pipes based on pipes taken as they come in the field? Can such a 

 formula show more than the new formula shows in Table 2 and on Plate VII? Except 

 in isolated cases this agrees within ±15 per cent with more than 250 observations on 

 52 pipes from 1 to 162 inches in diameter. Fiu:thermore, more than half of these pipes 

 are siphons on irrigation projects; that is, pipes inserted between open sections of canal 

 from which all manner of trash, silt, and rock ravelings may enter the pipe and retard 

 the flow. These are unavoidable conditions in the operation of such pipes, and the 

 author would not confine his tests to pipes under more nearly ideal conditions, even 

 if such might be chosen. These conditions must be anticipated by the designer and 

 some suitable factor of safety introduced as is suggested on page 66. 



The above is not offered by way of apology for the new formula but to make clear 

 the fact that the actual discharges from pipes designed on the basis of the new formula 

 are not to be expected to agree with the formula values exactly but only to a reasonable 

 degree. 



The author agrees with 'Mr. Williams and Mr. Noble that experimental errors, due 

 to erroneous assumptions of pipe diameter, may creep into the results. These errors 



1 Trans. Amer. Soc. Civ. Engin., vol. 51 (1903), p. 306. 



