Naval Architecture and Ship Building 125 



of the wave, but the majority of our calculations refer to the more or 

 less standard condition when the height of the wave is one-twentieth of 

 the length. The ship's structure as a whole is considered as a built-up 

 girder with Upper aad lower flanges and connecting webs. Knowing 

 the loading of the ship, it is possible, by methods which are tedious but 

 not difficult, to determine the bending moment of a vessel at any point 

 in any wave distribution and hence from the known characteristics of the 

 section of the vessel at that point we can calculate the stresses upon the 

 material o^ the hull structure. 



We come now to the crucial question, how nearly do these stresses 

 and strains which we obtain by these calculations approach what we 

 actually find in service? It is possible, with modern strain indicators, 

 to measure stresses of a ship in a seaway, but such experiments are ex- 

 ceedingly tedious and if properly done are expensive. In 1903, after a 

 destroyer had broken in two in Great Britain, numerous experiments were 

 made under the direction of Professor J. Harvard Biles in order to 

 obtain first-hand data upon stresses of ships at sea. It was found that 

 the results obtained were not inconsistent with those to be expected from 

 the standard methods of calculation, but they were not sufficient to 

 say that they extended to ships other than the particular vessel experi- 

 mented with. The real method which we rely upon in this connection 

 is a statistical and somewhat indirect one. Suppose, for instance, we 

 have a number of 5,000-ton vessels and find that, applying to them our 

 standard methods of computing stresses, the deck tensions vary from, 

 say 5tbns per square inch to 12 tons per square inch. If now we find 

 in long service that the vessels whose calculated stress is ten tons per 

 square inch and over show defective strength not necessarily by breaking 

 in two but by joints opening, rivets shearing, etc., while the vessels of 

 lower calculated stress give no trouble, we conclude that for such a vessel 

 we should ma^ke our plates so thick that the standard calculated stress 

 will not exceed eight or possibly nine tons per square inch, leaving some 

 margin below the danger point. It is essentially by this method that 

 we fix the strength of vessels to-day. Classification societies, such as 

 Lloyds, have been collecting information as to casualties, signs of weak- 

 ness, etc., in service for many years, and nearly fitty years ago scientific 

 papers in the British Institution of Naval Architects laid down in full 

 the principles which I have briefly outlined above. 



There is one point I may iTBention in this connection which shows that 

 our methods, adequate though they may be for practical purposes, still 

 leave something to be desired from the point of view of scientific prin- 

 ciples. If our methods were scientifically accurate they would show the 

 allowable stress per square inch the same in the case of large and small 



