]\gt ; The Determination of projectile if ) velocities and deviations of the axes are equal .
The mass of the shot , ) noment of inertia ; tilting couple ; hence the velocity of precession , which varies as the couple \mdash ; :moment of inertia , The curvature of the path due to the normal component of the resistance :molnentum of .
Radius of the helix \fnof ; g the velocity of preces. .
The radius of the helix of a -inch bullet would therefore ) of the order of inch , which would not be detectable on the screen records .
] On the Deteimi , tio7b of Fresnel 's Integrals .
By H. SHAXBY , B.Sc. ( Loud .
) , University College of South ales and lfonlnouthshire .
( Communicated by Principal E. .
Griffiths , eceived March 9 , \mdash ; Read Apri122 , 1909 .
) The ions J and ) , known as Fresnel 's integrals , have usually been evaluated by means of instance those of Cauchy and Knochenhauer .
The method of ation , by the summation of the areas contained between Jiven ordinates , the arc of the curve they cut off and the -axis , is readily applicable in this case , the quantity plotted as ordinate ainst .
as abscissa .
The area can then be determined between any yiven 1 Simpson 's Rule .
In practice , veyer , it is simpler and more accurate to apply Simpson 's directly to the calculated values of the ordinates , without a curve .
The curve has its zero values at , etc. , and is thus marked off into a series of loops of steadily decreasing area .
siulplest method of computation is to find the areas of these separate loops ( the first being only half a loop ) , in values of , from to 1 , 1 to 35 , , etc. The range for a particular loop will be termed its base-line .
In the subjoined tables the calculations have been made by each base-line into 10 parts and the ordinates corresponding to the points of division , for the first half-loop .
The area of any loop may then be written as , where is the length of the base-line and a factor depending upon the loop considered .
The .332 Mr. J. H. Shaxby .
On the phical [ Mar. 9 , values of are shown in Table I for the first six loops and for two other loops of higher order .
The table shows that , rapidly approaches a constant limiting value , .
Thus after a very few loops the area of a loop is at once given by the product of this constant and the length of the base-line .
The theoretical limit of is discussed in Section 4 .
The integration to any upper limit , where is an odd integer , is thus readily performed by adding to the integral for the first few loops ( say up to ) the value for the higher loops , where represents the quantity obtained by summing up the base-lines , attributing to them positive and negative values alternate ] ( because of the alternating signs of the loop-areas ) , and writing Table I exhibits also the areas of the first six loops , and so provides the data for such a calculation .
The values of ) can similarly be calculated for the loop-areas , where the base-lines range from to 23 , 2$ to 2 , 2 to , etc. , the values of for which These loop-areas and the corresponding values of are shown in Table II ; we again find the rapid approach of to the constant value Table I.\mdash ; Constants and Areas of Loops of Table II.\mdash ; Constants and Area of Loops of 1909 .
] lntegrals .
2 .
Integration can of course be performed to any upper limit by applying Simpson 's Rule to the portion of a loop between and , where is the odd integer less than in case of , the largest even integel ' less than in the case of .
A simpler method is given in Section 5 , for cases in which is not small .
Table III gives the values of ) for , etc. , 1 , calculated without further subdivision of the base-line of the first half loop , Gilbert 's values being added for comparison .
It is interesting to note , with a view to the availability of this procedure in cases where other methods may not be available , that such subdivision can give satisfactory results .
Table III .
Upper limit .
Gilbert 's table .
3 .
The values of and ) in the special case when and are not small and differ but little from each other , can be directly found as follows:\mdash ; and , approximately .
So , approximately , , whence .
( 1 ) Similarly , .
Mr. J. H. Shaxby .
On the Graphical [ Mar. 9 , 4 .
Special cases .
In the case when , where is any integer not too small for the approximation of Section 3 , the integral becomes where the positive sign refers to even values of , the negative to odd values .
If , in addition , , the value reduces to or , as in this ' case , to .
( 3 ) This expression therefore gives the area of the loop of the curve comprised between the limits and , and is numerically equal to of the , base-line of the loop .
( b ) .
A similar ation shows that if , the value becomes , where the positive sign is to be associated as before with even values of If we consider one loop , i.e. if , the integral becomes again 2 , showing that the value of is again 5 .
, where is the greatest odd integer less than From equations ( 1 ) and ( 3 ) computations may now be made rapidly ; thus , for example:\mdash ; .
The last quantity , so the integral is can be computed in a similar way .
the integer corresponding to being even instead of odd .
shows the values of the integral ) calculated by the above method for upper limits ranging from to by steps of 1909 .
] ofFresnel 's lnteyrals .
Table Upper limit 6 .
The method is also capable of the values of the rals between the limits and .
Thus ' ( 4 ) .
where being any If and can be estimated as by the method of nate summation .
of series by differential coefficients .
* Hence Boole , ' Finite Differences , ' p. .
90. .

