]\gt ; discontinuity be eased off , infinitely small wave-lengths may not be necessary .
In order to illustrate this question I commenced several years ago calcula- 3 tions relating to a very simple case .
These I have recently resumed , and !
although the results include no novelty of principle they may be worth putting upon record .
The case is that where the ordinate is constant between the limits for and outside those limits vanishes .
In general .
( 1 ) Lord Rayleigh .
The same set of values of Si up to Si ( 60 ) would serve also for the calculation of for and from to at intervals of It is hardly necessary to set this out in detaiL An inspection of the curves plotted from the above es shows the approximation towards discontinuity as increases .
That the curve remains undulatory is a consequence of the sudden stoppage of the integration at .
If we are content with a partial suppression only of the shorter , a much simpler solution is open to us .
We have only to introduce into ( 1 ) the factor , where is positive , arid to continue the integration up to .
In place of , we have .
( 9 ) The discontinuous expression corresponds , of course , to .
If is merely small , the discontinuity is eased off .
The foHowing are values of , calculated from ( 9 ) for ' As is evident from the form of ( 9 ) , falls continuously as increases , whatever may be the value of -i

