STEADILY ALTERNATING CURRENT ON A LONG LINE. 347 



of regular current transients at A. The oscillogram of is should 

 coincide with this line, if there were no errors in assumptions or 

 computation. The actual oscillographed is wave is shown by a 

 chain line in Fig. i6, and the deviations of this from the heavy line 

 are marked thereon at the successive crests. It will be seen that, 

 in the first wave, the discrepancy is small. The discrepancy in- 

 creases up to wave No. 3, when it reaches 17.5 per cent. The dis- 

 crepancy then dies away, almost to zero, in the last wave. No. 13. 



This discrepancy in current amplitude, reaching a maximum at 

 the third wave, is too large to be attributable to errors of interpreta- 

 tion in the oscillogram. It may be attributed therefore either to 



(i) small higher harmonics in the wave of impressed e.m.f . ; i.e., 

 to deviation from the sinusoidal form, with initial wave tran- 

 sients resulting therefrom, 



(2) imperfect regulation of impressed e.m.f., or initial deviation of 



Cs from the normal amplitude, with initiating wave transients 

 resulting therefrom, 



(3) lumpiness in the artificial hne, with corresponding initiating 



wave transients, 



(4) secondary reflections at A from the internal impedance of the 



alternator, not taken into account. 



The shares of the above possible sources of discrepancy in is have 

 not been determined. It is intended, however, to substitute a taper- 

 ing pair of smaller tt sections, for the first section at the A end of 

 the line, so as to diminish the lumpiness of the line at the generating 

 end, where the waves start. Ordinarily, the greatest discrepancies 

 between theory and observation are to be expected in the initiating 

 wave transients of current at the sending end. All errors are 

 exaggerated at this point. Considering the complexity of the con- 

 ditions, it is remarkable that the discrepancies have not been larger. 

 Of the considerable number of initiating wave transient oscil- 

 lograms which have been obtained on this and other artificial lines, 

 Fig. 9 is a fair sample, and not a specially selected case. It has 

 been analyzed to a greater extent, however, than any of the rest. 

 The conclusion to be drawn from this analysis seems to be sup- 



