334 
22 
Of course (38) will oscillate, since this conservation is only an approximate 
one. But after its average statistical behavior has been determined, any 
excessive deviation is an indication of some error of computation. Any 
clear trend of (38) with § is an indication of an inadequacy of the comput- 
ing setup, particularly if it is in the direction (decrease) and of the order 
of magnitude of the shock degradation of energy according to (21), (22). 
Here the numerical rate of degradation is the significant quantity, i.e, the 
degradation of energy while ¢ increases by |, i.e. during t units of time. 
By (21), (22) this is 
(39) A SCC ae Ne 
ur 
Hence oscillations and trends of (38) must be judged by comparing 
them in size with this A. ., 
(C) It is advisable to acquire some general routine regarding the 
adequacy of computing setups, by dealing first with some selected experimen- 
tal problems, i.e. simple problems in which the hydrodynamical solution is 
known. The comparison of the approximate, numerical solution with the 
(known) rigorous, hydrodynamical one will clearly be of considerable orient- 
ing value. The problems should be chosen in such a manner as to make it 
sure that shocks will develop, and preferably also other characteristic 
features of continuum hydrodynamics, such as Riemann rarefaction waves, wave 
reflections and intersections, etc. 
§16. The Ballistic Research Laboratory of the Ordnance Research 
Center, Aberdeen, Maryland, is carrying out explorations of the suitability 
of its punch-card equipment for certain computations of the type described. 
This work began in early March 1944 and is continuing at the present time, 
under the direction and following the setups of Mr. L. E. Cunningham of the 
laboratory. The author wishes to take this opportunity to express his 
