504 THE ROYAL SOCIETY OF CANADA 



iv. Where, under proper application, a process considered to 



HAVE been established BY EXPERIENCE AS CORRECT FAILS IN 

 ANY INSTANCE TO PRODUCE A CORRECT RESULT, THE PROCESS IS 

 THEREBY SCIENTIFICALLY CONDEMNED AND OUGHT TO BE ABAN- 

 DONED UNLESS THE FAILURE CAN BE DIFFERENTIATED, i.e., UN- 

 LESS A CONTROLLABLE CAUSE OF IT CAN BE LOCATED AND THE 

 PROCESS SO ADJUSTED AS TO PREVENT SIMILAR FAILURES, 



Illustrations: — (a) Where differentiation is possible: The bridge 

 of the Quebec Bridge and Railway Company over the St. Lawrence 

 river near the city of Quebec collapsed in course of erection on August 

 29, 1907. The span that collapsed was designed for more than twice 

 the length of any previously constructed in America. In the subse- 

 quent official inquiry it was ascertained that a variety of causes, in- 

 cluding the excessive unit stresses of the entire structure, tended to 

 induce the disaster; the actual point of failure, however, was in the 

 lower chords of the anchor arm near the main pier, the failure being 

 due to the weakness of these compression members or columns, and 

 especially to insufficient latticing of the chords. The defective design 

 of the lattice was not due to a lack of common professional knowledge 

 on the part of the responsible engineers. These engineers, who were 

 prominent members of their profession in America, depended neces- 

 sarily, in designing a span of the adopted length, on the experience 

 of themselves and others gained on much smaller bridges. With 

 respect to the lattice of the columns or posts, they exercised their 

 judgment in points which are necessarily left to the individual judg- 

 ment of the engineer. This judgment proved faulty; but the engin- 

 eering profession, though it has learned much from their mistake, 

 is not yet in a position to determine the percentage of their error. 

 Until this determination has been made, it will be impossible to design 

 latticing for such lengthy spans that will be unquestionably safe 

 and not unnecessarily heavy. Meanwhile, however, it is not necessary 

 to forego their construction because, while the professional knowledge 

 of the day is not sufficient to design such structures economically, a 

 bridge of the adopted span that will be unquestionably safe can be 

 built by using a considerably larger amount of metal than might be 

 required if this knowledge were more exact. In other words, with 

 respect to lengthy spans, the engineering process of bridge construc- 

 tion which hitherto proved in experience to be correct, is not scienti- 

 fically condemned and need not be abandoned on account of the 

 Quebec failure, because that failure has been differentiated, i.e., a 

 controllable cause of it was located in the weakness of the columns or 

 posts and by using an excess of metal in these posts that process has 



