553 



From these distribution curves the standcird deviation and probable 

 error of a single observation may be calculated. These are given in 

 Table XI. 



Table XI. Deviation measures for 212 damaged diaphragms 



(d) Test of Hopkinson's scaling rule . In order to test the well 



known Hopkinson scaling ruleiz/, half -scale models of the regxilar UML 

 diaphragm gauge were constructed. Each linear dimension of the regular 

 gauge was carefully reduced by one-half. To insure that the properties 

 of the half- scale diaphragms would be the same as those of the full 

 size diaphragms, we used copper diaphragms einnealed under the same con- 

 ditions and at the same time. 20 BS gauge (thickness .032 in.) was used 

 in the small gaviges and l**- BS gauge (thickness .O65 in.) was used in the 

 regtilar gauges. To eliminate the necessity of scaling boosters and sub- 

 sequent difficulties of detonating cast charges, loose tetryl charges 

 were used thro\aghout. In each case the container used for the charge 

 with the small gauges was scaled from the larger container (except for 

 wall thickness) and the weight of explosive was I/8 that of the larger 

 charge . 



The results listed in Table XII indicated that the damages scaled 

 on the average to within 2%, the largest deviation being 5^. The shapes 

 of the pairs of diaphragms had the same appescrance. If there was any 

 change in the strength of the copper diaphragm with rate of strain in 

 the raaige of rates encountered here, it must have been masked by com- 

 pensating errors. Any such "speed effect" shovild cause a deviation from 

 Hopkinson's rule. It should be noted that part of the shots were with 

 wooden frames suid the others with steel rings. When the steel ring 

 support was used, copper diaphragms showed the same weight suid distance 

 exponents as previously obtained for copper diaphragms using wooden 

 frames {O.k and 0.8 respectively). 



1^ / "the damage inflicted on a given sti^cture by a given charge at a 

 given distance will be reproduced to scale if the linear dimensions of 

 the charge and the structure and the distance between them are aJ.1 in- 

 creased or decreased in the same ratio"; RE l'4-2/l9 Submarine explosions , 

 p. 18, H. W. Hilliar. 



