ABSTRACTS OF TECHNICAL PAPERS 813 



determination and application of methods of allaying it. The trouble 

 was experienced mainly on the Pacific Coast where, although Douglas 

 fir conduit was introduced about 1911, the first case of corrosion which 

 could definitely be ascribed to the creosoted conduit did not occur till 

 1921. 



A search for the cause of the trouble led to making systematic anal- 

 yses of the air present in the conduit and these analyses revealed the 

 presence of acetic acid in sufficient amount to account for the cor- 

 rosion in the presence of carbon dioxide which was also shown to be 

 present. 



After much experimenting a method was developed to stop the cor- 

 rosion by pumping ammonia into the ducts. Results have been very 

 satisfactory and seem to indicate that a single treatment is sufficient. 



Small Samples — New Experimental Results}'' W. A. Shewhart and 

 F. W. Winters. This article reviews briefly the Theory of Errors of 

 Averages, paying particular attention to some of the most recent 

 work in connection with small samples. New empirical results are 

 presented showing the advantage that arises from the use of the latest 

 error theory and pointing out the effect of the limitations imposed upon 

 it. The information contained in this paper indicates that further 

 theoretical studies ate necessary in order that the application of small 

 sample theory may give more accurate solutions to the problems that 

 arise in practice. 



" Journal of American Statistical Association, New Series, No. 162 (Vol. XXIII), 

 pp. 144-153, June, 1928. 



