A rough estimate of the accuracy of the force computations can be 

 obtained from two sets of computations in which only the time step is 

 different. One such numerical test was conducted for this problem with the 

 frequency parameter o^b/g = 1.7 and the amplitude of the motion such that 

 hg/b = 0.3. The two time steps used were At = 0.02 and At = 0.01. After 

 about t = 2, the average difference in the computed force was less than one 

 percent. Thus the error seems to be quite small. 



Various frequencies of forced motion have been considered for the 

 amplitude corresponding to hg/b = 0.3. Predictions of the coefficients for 

 second-order theory were made and compared with those of Papanikolaou and 

 Nowacki [18], Potash [19], and Lee [17]. The results are presented in Figures 

 12 and 13. The computed first-order coefficients agree well with the 

 previously computed results. The computed second-order phase angle agrees 

 well with the results of Papanikolaou and Nowacki. The magnitude of the 

 sinkage (the third term in Equation (28)) agrees well with the previously 

 computed results, but the magnitude of the oscillatory part of the second- 

 order force does not agree so well. This is a reflection of the relative 

 error in the nonlinear force when the first-order force, accounting for most 

 of the force, has been subtracted. It is doubtful that there is more than 

 about one digit of accuracy in the computed second-order results. To obtain 

 these results, the fluid motion resulting from about two periods of forced 

 harmonic oscillation was computed, and the force for the last period of the 

 motion was decomposed into its Fourier components. 



23 



