]\gt ; On Nature of the Diffraction Figures due to the Heliometer .
303 light in the image had been undertaken by Schnauder .
The only other paper on the subject was by Scheiner and Hirayama .
It consisted chiefly of plates of photographs of diffraction images formed by various apertures , including the semicircle , and a summary the paper by Bruns already .
There are also notes and a definite statement that the calcu- lations initiated by Prof. Bruns had not been completed .
3 .
I decided to attack anew the problem of computing the diffraction figure , hoping by the help of modern mechanical- integrators to be more successful .
The formulae developed by Bruns give ths light intensity , , in terms of two components and , such that where , and ; and being the ray and angle polar co-ordinates of any point of the diffraction image referred to the centre of the image as pole , and a line parallelto the diameter of the aperture as imtial line , and , a constant such that , if the intensity at the centre of the image be taken as unity , we must put The values of are obtained without difficulty from any table of Bessel functions for a somewhat limited range of values of , but the values of depend on the evaluation of the integral Three methods gest themselves , namely ( 1 ) to make this integral depend on other tabulated rrals , ( 2 ) to expand in series , and ( 3 ) to evaluate by mechanical integration .
It was not found possible to use the first method , and so only methods ( 2 ) and ( 3 ) were available .
Since , it is clear that the choice of method depends largely upon the range of values of that are to be considered .
Now in the portion of the diff'raction figure actually drawn ranges from to .
For the central region from to expansion by Taylor 's theorem was used , and for the remaining portion , where expansions of the integral converged with extreme slowness , mechanical integration was adopted .
4 .
Evaluation by expansion , using Taylor 's theorem .
With the usual notation , let be the value of the function I and the value of the nth derived function when is put equal to after obtaining the derived function .
Let Mr. P. F. Everitt .
On of the [ Sept. 11 , then ; ; or ; Differentiating again , ; .
Proceeding , whence As we proceed further we shall always , in the case of the derived functions of even order , have on the right-hand side , as factors under the integral sign , an even power of and also a , consequently For the odd derived function , the law is best expressed as follows:\mdash ; whence Six terms in the expansion , i.e. proceeding to , were found to be sufficient.for the range of fairly small 's dealt with by this method .
The values of the integrals were obtained by expressing the powers of in terms of cosines of multiples of and then integrating .
The values so obtained were\mdash ; Mr. P. F. Everitt .
On the Nature of the [ Sept. 11 , The work was arranged in parallel columns , each column for one value of and each sheet containing all the work for one value of .
For the preliminary calculations , values of .
were taken at intervals of , the values of so obtained were plotted on squared paper , and the maxima and minima ( which will be separately discussed ) inserted .
From this graph it was evident what further points were required in order to draw the curve of accurately .
This rough plotting on squared paper served as a very useful check against errors in the work of calculation .
The arrangement is shown in Table III .
The positions of the maxima and minima were calculated by a different method .
Table III.\mdash ; Calculation of a in this table denotos that the numerical quantity of which the logarithm is given occurs with a negative sign .
Since and then .
Hence , for anyvalue of zero , not being zero , and have their zero values simultaneously ; ' consequently , instead of considering we may take .
We have then\mdash ; 1909 .
] ' or Diffraction Figures due to the Heliometer .
and the roots of this equation give the values of , and therefore of , for which is a maximum or minimum .
The roots were found by successive approximation .
Put then Now expand the left-hand side by Taylor 's theorem and the right by the binomial , second and higher powers of Then ; Therefore 2 .
This formula was not very often required , as is easily seen from the considerations .
Counting the maxima and minima from backwards to , there is only a limited number of them for any given value of ; moreover , these maxima and minima occur precisely the same values of and , when takes up a new value , the only with regard to them being that changes its value in accordance with The values of the maxima and minima may be calculated by the same method as the other yalues of , or more ] as follows:\mdash ; ; therefore ) .
All the data required for the plotting of the curves of for the different values of being now obtained , the curves were plotted on large , thin , smooth-surfaced* cards in rectangular co-ordinates by means of a Coradi co-ordinatograph , and the curves drawn in with the weights and splines used in naval architecture .
The integration was then performed by means of a Coradi integraph , which draws the raph of the integrated function .
The cards were again placed under the raph , and * Because both co-ordinatograph and integraph work best on such cards .
Mr. P. F. Everitt .
On the Nature of the [ Sept. 11 , the ordinates ruled for successive values of , increasing by steps of The photograph of such a card is shown on a much reduced scale in Diagram 1 .
DIAGRAM 1.\mdash ; Mechanical Integration of the -Integral .
( Reduced to 1/ 6 linear of original drawing .
) The values of first taken were those giving the maxima and minima of the ring system of the circular aperture , and afterwards a sufficient number of intermediate values were taken to give an accurate curve when was .
plotted against .
The values of the ordinates of the integrated curve were scaled off and tabulated .
For the angles and the values of I were obtained by both methods , and these values were compared .
The values are given in Table These show a maximum difference of 1 in 457 , or about 2 per mille .
The greater the value of , however , the greater the inaccuracy of the mechanical integration , but the increase is slow ; but , on the other hand , we have also \amp ; less accurate result from the less convergent series .
1909 .
] Diffraction Figures clue to the Heliometer .
The value of was obtained by both methods in four cases and the results are given in Table .
It is at once evident that the mechanical ration gives accurate results , the maximum error barely one unit in the fourth place of decimals .
The above method , as I have said , was available up to , beyond which the 's are not yet tabulated .
As this range does not carry us beyond the third bright spot , and it was desired to go further than this , another method was necessary .
After several futile attempts , such a method was found and has proved itself very useful .
It would , of course , have been possible to have computed all the 's required ( a very large number ) , but the labour would have been excessive and the following method was adopted to lessen the work : ( sin ) Differentiate both sides with regard to but therefore Therefore ) Mr. P. F. Everitt .
On the of the [ Sept. 11 , this value in we have ; or , putting The determination of resolves itself , therefore , into the evaluation of this integral .
The method adopted may be called a semi-graphic one and is identical with that used independently by J. H. Shaxby and described by him in a paper read before the Royal Society on Apri122 , 1909 .
The method had been found , tested and definitely adopted for this work at the end of the Lent term , and before I was aware by Mr. Shaxby 's paper that he also had discovered its advantages .
The method depends on the property possessed by this class of curves that , after cutting the axis a third time , the areas of all the other sections are almost identical with the areas of sine curves of the same altitude on the bases formed by the various points of intersection of the original curve with the axis .
This is seen ) the values given for and in Table Table \mdash ; Areas of Sections .
The horizontal line drawn through the above table marks where the curve cuts the axis for the third time .
Beyond this point the calculated areas are practically identical with the planimetered areas of the curve .
320 Mr. P. F. Everitt .
DIAGRAM 5.\mdash ; Sketch of Apparatus employed .
On the Nature of the [ Sept. 11 , a scale .
By this means the circular aperture is moved over the until half of the circle is covered .
The adjustment to the exact semicircle was made by means of a Hilger travelling microscope .
This aperture was designed so that , after calibration of the micrometer screw , the diffraction fringes of any circular segment could be accurately found .
The remainder of the apparatus used ind the general arrangement is shown in Diagram .
The equivalent focus of the telephoto-camera was about 19 feet .
In consequence of the optical peculiarities of such a system it was necessary , if this focus was to be retained , to have the point source of at a great distance , which , unfortunately , could not be done .
A collimator was accordingly introduced ; and , as the of the pinhole source of light was magnified in the ratio of the foci of camera and collimator , a small of the pinhole ) roduced by an eye-piece was used , the only disadvantage being a largely increased exposure .
The exposure required to obtain the photograph reproduced ( Plate 3 ) was three hours .
The greatest difficulty met with arose from the object-glass of the telescope having a number of waves of unequal refraction in the lass , and the parts so affected had to be most carefully avoided , as otherwise the diffraction would have been adyersely affected .
Mr. P. F. Everitt .
On the Nature of the [ Sept. 11 , The values for the major axis were obtained from the intensity curve shown in Diagram 4 , while those for the minor axis are , of course , the values for the diffraction image due to a circIe .
There is an interesting point about these values .
In the case of the minor axis ( or circle ) each maximum lies nearer to the minimum of its own order than to that of the next higher order , but this tendency of the maxima decreases as the order increases , and ultimately vanishes .
In the case of the major axis , the same tendency of the maxima is present ; but it appears that the tendency , although becoming smaller , does so far less rapidly .
Turning now to the contour diagram ( fig. 2 ) , and comparing it with the photograph ( Plate 3 ) , the agreement is perfect ; and while the , photograph makes reference to the contour diagram easier , the contour diag'ram emphasises and extends the details of the raph .
From the contour diagram we see that the central spot is not really an ellipse , but only an approximation to one .
Calling the five contour lines shown in the central spot counting from the centre outwards , we find , on measuring their diameters along the axes of the figure:\mdash ; We see that there is , on the whole , an increase in the ratio of the diameters , so that the shape of the central spot varies with the exposure when taking a photograph , and also with the intensity of the light when observed visually .
The deviations from the steady increase in the ratio which are shown by and may be quite reasonably explained by the influence of the first bright ray , which should increase the ratio very slightly in the case of , and of the first dark ray , which appears to decrease the ratio somewhat in the case of C. In connection with this , the way the contour line we have just called is pulled out of shape is very significant .
10 .
General description of the diffraction image .
The image is , as has been already mentioned , symmetrical about two perpendicular axes .
There is a central approximately oval spot , which is roughly a little less than twice as long as it is broad .
This central spot is bisected along its major axis by a long bright ray , which fades away as it recedes from the centre .
Around the spot and following its outline are a 1909 .
] Diffraction Figures due to the Heliometer .
series of alternate dark and bright rings , which , at the minor axis , are identical with those of the diffraction image of a circular aperture , and , along the major axis , terminate at an in a series of bright spots .
Further , these rings are crossed and modified by a series of alternate bright and dark rays of approximately hyperbolic shape , having the major axis of the figure for their geometric axis , which they cut at its maxima and minima , with the exception that the first one of the series passes through a point in the central bright spot .
The whole action image presents an exceedingly beautiful appearance , which is , unfortumately , only very inadequately represented by photography , but which is very easily studied with the double-star apparatus when monochromatic light is used and the star is examined in the equatorial through the heliometer lens .
Conclusions .
( i ) The complex diffraction fringes due to a semicircular aperture have been for the first time theoretically determined , and actual measurements on the fringes themselves shown to be in close accordance with the calculated results .
The absolute agreement of the computed contours with the photographs forms another item in the evidence , showing the old undulatory theory of light suffices to describe very accurately diffraction phenomena .
( ii ) A problem which has attracted the attention of more than one astronomer and mathematician but had not yet been overcome , has been solved , and the exact nature of heliometer images numerically determined .
( iii ) The solution of the problem has illustrated the limits to the practical utility of Bessel 's functions ; the mechanical integration of 24 indefinite and 17 definite integrals has indicated the manner in which graphical methods can be used to supplement numerical analysis , and has contributed to the solution of a physical and astronomical problem of considerable interest , which had hitherto proved unattainable by the usual methods of evaluation .
BIBLIOGRAPHY .
Bessel , 'Astronomische LTntersuchungen , ' Bd .
I , 1841 ; 'Astronomische Nachrichten , ' Bd .
VIII , S. 411\mdash ; 426 .
Bruns , " " Uber die Beugungsfiguren des Heliometer-Objectivs , \ldquo ; 'Astron .
Nachr Bd .
CIV , No. 2473 .
H. Scheiner and a Hirayama , " " Photographische Aufnahmen Fraunhofer'scher Beugungsfiguren ' Abhandl . .
Wiss .
in Berlin , ' 1894 .
J. H. Shaxby , " " On the Graphical Determination of Fresnel 's Integrals ' Roy .
Soc. PIOC A , vol. 82 , No. A 555 .
Gray and Mathews , 'Treatise on Bessel Functions and their Application to Physics , ' London , 1895 .

