]\gt ; INDEX WALXER , G. T , \mdash ; On Boon)erangs , Phil. , 1897 , vol. pp. Boomerangs , aooount of ; air-pressure on ; traJeotories of .
Walker , , T. Phil. Trans. , , 1897 , vol. 19o , .
23-41 .
G. T. WALKER , M.A. , B.Sc. , of College , ) .
by J. J. Reccived Mnrch 15 , \mdash ; Read THE att:empts that have hitherto been made explain the of a boomeranlg have in general been of a somewhat nature .
Exception must be made in the case of such papers as those of STILLE , ' ' Versuche und Rechnullge sir Bestimmung der Bahnen des Bumel.angs\ldquo ; DORFF , ' malen der Physik , ' Bd .
147 , lS72 ) , and of EDMUND GERLACfI , " " gewisser Bewegungsformen geworfener Scheiben aus dem ( ' des Deutschen Vereins sir orderuug der Luftschifffahrt , ' Heft 3 , In the latter , which is the most iceable contribution to the ubject with which I am , the author gives an explanation in general terms of some of the effects of the air-resistance upon a symmetrical boomerang : he introduces , ever .
no analytical treatment of the dynamics of the rotating body and neglects entirely all consequences of the important deviations from symmetry which I have subsequently described as " " twisting\ldquo ; and " " rounding Without one of these a return flight is , I believe , impossible .
For am account of the native Australian weapons , in particular those of reference should be made to the very complete descriptions given in BROUGH SMYTH'S book , The Aborigines of Victoria , ' vol. , pp. ; shorter notices are to be found in books of travel , such as that of KARL LUMHOLTZ , ' Among Cannibals , ' p. 50 .
Boomerangs may at the outset be divided into two classes\mdash ; re ning and nonreturning ; it is rather on weapons of the latter of these types that the natiyes of Australia rely when engaged in war or the chase .
A typical returning ( see fig. 1 ) resembles in general outline an arc of a hyperbola , and is about 80 in length measured along the curve .
At the centre , where the of cross section ( fig. ) are greatest , the width is about 7 centims. , and the thickness 1 centim. ; these dimensions become smaller as the ends are ched .
As a rule two properties are present .
In the first place , the transverse section at any point would show that one surface possesses distinctly greater curvature than the other ; secondly , the arms of the implement must be slightly twisted coincidence with the plane through each of them ) after the fashion of the blades 7.9.97 scre opeller or a windmill .
The direction of the twist is such that rotation about a normal to the plane to set up linear velocity of the boomerang direction the vector representinoa .
that .
These two peculiarities will in future be referred to as the " " rounding\ldquo ; and the " " twisting A weapon of this type is thrown in a horizontal direction in such a way as to impart considerable rotation in the vertical plane containing its initial direction of motion ; the more convex surface is towards the thrower .
The plane of rotation leans slowly over to the right ( i.e. , the vector representing the spin begins to point slightly upwards ) and the path curls to the left .
The projectile proceeds to describe a loop whose longer diameter is about fifty yards ; it gradually rises until it reaches a height which is usually about thirty feet the ground , travels horizontally for a time , and then gradually sinks to the earth .
The change in the angular motion has throughout the flight continued unaltered in character ; the inclination of the plane of ation to the izon has steadily diminished a right angle to zero , and the axis of the spin has veered continua ] lie to the left ( as seen from above ) in such a manner that as long as the linear velocity remains large , the angle between the direction of motion and the plane of rotation is small .
In ttJe accompanying diagram ( fi gs .
2 , 3 ) a plan and elevation of this , the simplest form of path , is given .
An attempt is made to indicate the inclination of the axis of rotation by representing at intervals the projection of a line of constant length drawn that axis .
If it be not desired to make so large a loop as that described , it is fairly easy to get the boomerang to describe a circle of thirty-five in , without ever rising to more than twelve feet from the earth .
In the more complicated paths , as long as the velocity remains considerable , the manner in which the plane of rotation and the direction of motion change is precisely the same as in the simpler cases ; it is the rates of change that differ .
The graceful gyrations that a boomerang performs on its downward course , if the linear velocity dies out while it is high in the air , present little or nothing that is new in principle .
It is in the explanation of the earlier motion that ] ) problem really lies , and the observation of actual flights makes it clear that their character is deducible when the MR. G. .
WALKER ON BOOMERANGS .
two components of angular velocity noted subsequently , whose axes lie in the plane of the boomerang , are determined .
The flight may be regarded as a case of steady motion of which the circumstances gradually vary .
It is only with very badly made instruments that small oscillations are at times perceptible ; with ordinary boomerangs , the accident of grazing the ground or meeting a sudden pnff of wind will not cause visible vibrations .
Fig. 2 .
Plan .
The scale of this and the following diagrams is 1 : 1000 , or 28 yards to 1 inch , approximately .
Fig. 3 .
Elevation upon a vertical plane through Let the containing the arms of the boomerang ( in future called the primary plane ) be taken as that of , with the centre of gravity as the origin , and the projection upon this plane of the resultant velocity as OX ; is drawn on the more convex side .
If then the rectangular components of linear and angular velocity of the body be , and , it may be observed that is always smadl compared with , and compared with .
Throughout the motion is positive , negative , and positive .
The time of flight is about nine seconds , and the greatest distance fifty yards ; the mean values of and may be estimated at one-sixth and minus one-third respectively , while in C.G.S. units is two thousand and is thirty .
The angular velocity of the axes is small and positive throughout the motion , except near the conclusion , when it sometimes vanishes and becomes negative .
theoret , ical purposes I have regarded the body as replaced by one of extremely thin material with the same general shape and twist ; the transverse section will be a circular arc with its convex surface on the same side as the more rounded surface of the wooden weapon .
Experiment shows that if a thin rectangular plate be advancing with velocity in a direction that is inclined at a small angle to its plane , the air-pressure produces a VOL. CXC.\mdash ; A. force and a couple about the centre ; the mean normal pressure per unit area be delloted by , where is a constant that depends on the proportions of the rectamd is a number has been often assumed to be equal to 2 .
In his " " Experiments in Aerodynamics\ldquo ; ( ' Smithsonian Contributions to Knowledge , ' 1891 ) , S. P. LANGLEY makes this assumption , but if from Table XIV .
the value of be calculated by comparing the soaring yelocities of planes , weighing 250 and 1,000 grms. in clinations of ( the smallest inclination quoted ) , it proves to be ; for square planes of the same weight , inclined at , the index is .
From comparison of the cases of inclination of and of Table XII .
, the value of may be deduced .
In the course of the following analysis it will be seen that progress is attended with extreme difficulty unless , and inasmuch as the constant is ' disposal , we shall be justified in taking choosing so as to agree with .
LANGLEY'S experiments at the mean value of the velocity under discussion .
Any error introduced by an incorrect value of will be quantitative rather than In addition to the uniform pressure acting on the rectangular plate , there will be a couple whose amount may be as per unit area , being a constant depending on the dimensions of the plate .
* The assumption of a velocity potential would lead to the while is suggested by the previous assumption .
In order to subsequent proceedings we shall choose the smaller value and deduce the value of from LANULEY'S experiments .
We have to consider the effect of the air on the slightly distorted thin surface which rsents the boomerang , ctnd in order to surmount the difficulties introduced by the fact that the velocities at diflerent points vary , as well as the directions of the normals to the surface , we are driven to make some hypothesis .
Now the effect of the air-pressure upon a plane surface uniform motion may be obtained by integrating over it , provided that we regard the effect due to ny small portion as proportional to the area of that portion .
We thelefore assume , as a first approximation , that the contribution from any ement of the distorted surface , is the same as if the rest of the surface were in the same plane as the element and had the same velocity ; that this assumption , in the of simple distortions , leads to results of the right character , is easily verified .
The determination of depends on the fact th if the width of an arm measured in the direction of the velocity of the point in question be , and if stand for the * See Natul'al Philosophy , ' S 325 .
The existelloe of this couple is often stated in the form that the resultant thrust on the plate does not act at the centre of figure .
LANGLEY finds chap .
viii .
, that in the case of a square plate the point of application of the resultant pressure , when does not exceed ten degreeH , is at a distance from the centre of figure equal to about one-sixth of the length of the side .
He quotes SSEL and KUMMER as having obtained a fifth and a sixth respectively the value of this ratio .
LAMH'S ' Hydrodynamics , ' p. 185( ; 3 ) ; BASSET'S ' vol. 1 , S 190 .
MR. G. T. ON BOOMERANGS .
ratio 1 : 5 or 1 : 6 as deduced from experiment , then couples and should be the same .
This would give a value of varying from point to point .
We therefore , for convenience , treat as constant , and give it the nitude corresponding to the mean value of It must be realised at the outset that the following analysis does not claim to be more than a first approximation , in which the quantities neglected may be of tenth of the.magnitude of those retained .
Our knowledge of the laws of the resistance of the air is not at present great enough for accurate results to be attainable , and I have accordingly not hesitated to neglect small terms in order to effect a material simplication in the mathematical analysis .
It may appear that such processes reduce the method to little more qualitative one , but though much may be done by qualitative methods applied to this subject , all the chief terms may be to their source without the use of algebraical symbols , yet , as wil ] soon becorne clear , the effects of the forces in action are conflicting .
It is therefore necessary , in order to obtain results which are qualitatively right , to adopt methods which , although not accurate , have at any rate approach to quantitative correctness .
We now take axes fixed in the body , 1 and being along and perpendicular to the axis of symmetry in the primary plane .
Fig. 4 .
If the velocity at any point have components \ldquo ; , and the direction cosines of the normal on the convex side there be , then the normal pressure in the direction will be , where are small quantities and The couple per unit area will have moment about an axis whose irection cosines are each divided by a quantity differing from by squares of small quantities .
Hence , if , the normal force at any point will be , and the component couples where squares of small quantities are omitted .
Now if , due to the " " rounding a transverse line ( fig. 4 ) through any point be a circular arc of radius , and if its middle point have co-ordinates , then , QP as measured the inward normal QN by , the direction cosines of the normal at will be .
In addition to this the line will , owing to the twisting , be turned about the tangent at through an angle which may be taken as where is a constant ' length .
The superposition of this small distortion on the former will add to the direction cosines terms Let the linear and ocities of the body referred to the axes 1 , 2 , 3 be where is small compared with and compared with .
The component velocities of will be The resultant force due to the pressures will have negligible components parallel to the axes 1 , 2 , and the force along the third axis is the integral over the surface of MR. G. T. WALKER ON BOOMERAN S. This may be regarded as the sum of three forces : ( 1 ) the force which is exerted on a boomerang without distortion , ( 2 ) due to the rounding , ( 3 ) due to the twisting .
Denoting by brackets the } ) eration of taking the mean value over the area of the boomerang , we shall adopt the notation sin Then for a plane surface we find at once where is the area , and terms in have been omitted since they are multiplied by the small terms The rounding produces a force Due to the twisting we have The couple axis will be the integral over the surface of ( lu ?
As before this may be divided into three portions , of which the first , on a plane boonlerang , is xv \mdash ; in which terms have been omitted as before .
Therefore , Sw while MR. G. T. WALKER ON BOOMERANGS .
The couple about the axis is the leading to \mdash ; 2 The equations of angular motion are Neglecting the product , we see that may be replaced by , a constant .
Also , for a thin flat body , is sensibly equal to , and if be the mass per unit area , we have so that our equations become .
We shall of all discuss the motion of undistorted boomerang free from the action of gravity .
If we revert to our formel axes OX , OY , , of which OX is the projection on the primary plane of direction of motion , we shall obtain as the equations of translation , Hence , neglecting squares , is constant , and , the angular velocity of the axes , is .
Thus we are justified in replacing by .
We shall then have . . . . . .
( 1 ) , and on multiplying the former rotation equations by and adding , we get say , Now - .
Similarly together with terms whose coefficients involve circular functions of Also .
MR. G. T. WALKER ON BOOMERANGS .
If all the terms be collected on same side there three equations of the form in which the coefficients ma contain the operator , or circular functions of the time .
The equations giving the motion under gravity of a with the two distortions will differ from these in three ways .
First of all , if be the direction cosines of a Iine drawn iCally downwards , the equations of translation will be bearing to the same ] to From the second equation ' and from the first Now in numerical value is comparable with , and with 30 , that the additional terms in ( 1 ) due to the consideration of will be of gligible m ude .
Again , is small when the path is nearly in a horizontal plane ; hence , in considering the steady motion corresponding to a particular portion of he path , the change in need not trouble us .
When the forces due to the distortio s\mdash ; and these wil ] not involve , w\mdash ; are introduced , we obtain ' where is the angle between the axis of revolution and the upward vertical , and being small , we shall treat in the third equation as constant .
We next regard the right-band sides of these equations as expanded in the form VOL. CXC .
3-1 MR. G. T. ON BOOMERANGS .
in which . . .
are constan ts .
On vnlnes of , we find Our equations may now be satisfied by the infinite series . .
ni . . .
, since the ratio of the coefiicients of and to those of and to be ultimately comparable with If we adopt the notation U- the on the left-hand sides becotlc -A .
WALKLR ON } S. Hence the substitution of the series for gives where linear functions of the coefficients contain or , but not the nnmerically more portant quantity , as a factol In order , then , to obtain a steady motion about which Ininute oscillations are on , we no , lecb the terms , in our approximation , This is to taking two points on the path at an interval corresponding to a number of complete revolutions ( say twelve ) , and asserting that the angular change in of rotation is that due to the non-periodic portion or mean of the couples in action dn that period .
For steady motion to be possible it is necessary that the values of by the equations shall ) small .
inserting values , it appears that is condition of to be large enough to give the determinant a positive If is small , th tion i unstable unless values , When the value of is 26 , and when armF } are at angles , stability is secured when These values are rather larger than those found necessary in practice , but their mutual relations are .
The first time that a beginner attempts it , he a boomerang whose arms are at righ angles travel steadily , but the more the * ON RANGS .
, the more difficult is the owing of the implement , and when or upwards , the material of which the boomerang is made is light , it against a wind requires skil ] of a order .
The values of the constants .
have been ated for boomerangs whose arms are 36 centims. in lengtl ] and 5 tims .
in width , the mass per unit area being five-eighths of a gramme .
Wllen these constants are substituted in the equations of steady motion , it is found that for a boomerang whose arms at cight , correspondirJg to the values are the velocities ( 3 ) .
\mdash ; 1600 If we make , the values of , given by the equations are too large ; this is due to the fact that the limit of stability is not sufficiently exceeded .
If the value of be taken as 5 instead of 7 ( these being estimated inferior and superior limits of corresponding to , and ) there appear \mdash ; \mdash ; 1400 ( 4 ) .
The velocities ding to a larger spin are ( 5 ) .
MR. G. T. ON BOOME R S. It is to these Stllts with those belonging to a boomerang whose arms include Thus , taking we obtain 3.8 2000 .
1 ( 6 ) , while second nate of , namely 5 , yields 4.6 2300 \mdash ; 3 .
( 7 ) .
The values of and ctice are usually comparable with 20 and 800 respectively .
That the form of the equations is correct , at any rate as regards first approximation , is confirmed by the experience gained in making and throwing upwards of seventy boomerangs of different weights , shapes , and sizes .
If , for example , one of these does not curl sharply enough to the left ( i.e. , is negative , but not numerically large enough ) , it is found that increasing the twist diminishing will produce the desired effect .
A furbher result will be an increase in and a lent tendency to ' ' sky this may be corrected by making the difference of curvature of the two surfaces more pronounced ; a diminution in will thus bring about a diminution of Some of these implements were made with the express object of verifying particular terms .
If there be no twist , and , while is not extremely large , is negative and positive ; if , on the other is infinite , but is finite and positive , is positive and is negative .
From experiments made in this manner , with a somewhat smaller spin than that assumed above , I have deduced the formulae ( 8 ) , .
G. T. WALKER ON BOOM } in , owing to the experimental difficulties , the numerical of bhe coefficients ma be regarded as lacking in accuracy ; they may however be elied upon as , at any , of the correct order of nitude .
Of I have nothing except that it does not exceed 500 , and is probably smaller and negative .
A comparison of the theoretical results ( 3 ) and ( 4 ) , or ( 6 ) and ( 7 ) , obtained with different data for will show that he formulae , calculaf , ed , must be looked upon as giving only a rough estimate of the motion regarded quantitatively ; but , in spite of the calculated vahle of being excessive ( between 600 and 1200 when ) , it will be seen that the discrepancies are of the kind that might be anticipated , and the theoretical equations are qualitative]y consistent with the experimental results given in ( 8 ) .
Another piece of evidence is that furnished by nonreturning .
If it be desired to make an efficient missile that shall travel in as straight a path as possible , it is natural to manufacture , boomerang without twist and the curvature of the two faces the same .
It is this form that many of the cruder Australian weapons possess .
Experiment and theory alike show , however , that if initiaJly have a positive value less than a right angle ( i.e. , the natural method of throwing be adopted ) , then will be positive and will be negative as long as is than angle : when the of rotation has reached and passed through the horizontal position remains negative .
The of the path is indicated in fig. 5 , and it will be seen that it is far from straight .
Fig. 5 .
A path in one vertical plane ] be secured by wing an Yndistorted weapon with its plane of rotatio1 ] accurately vertical ; the least inclination , however , would grow , and the plane of initial motion be departed from ; in any case , except for the reduction in the resistance of the air , the path in a vertical plane would yield no greater range than would be afforded by a spherical missile of the same weight .
We might attain the same end by igning the shape so as to make small when throwing the boomerang with its plane approximately horizontal .
In that case the plane would remain horizontal , and the axis OX in it would soon be pointing in a direction slightly above the tangent to the path ; a much longer flight would then be maintained , as the effect of gravity would be balanced by the upward of the air on the lower surface of the projectile .
It is interesting to notice that this is the method that experience has taught the blacks to adopt .
Their best non-returning weapons always have strongly developed positive rounding ( the more curved rface is uppermost when ) often a MR. G. .
WALKER ON S. small negative twist ; examiuation of the equations will show that these distortions will combine to produce the required results .
An estimate of the efficiency of the shape may be made from the fact as far as my experience goes , boomerano of this type may be thrown more than twice as far as a spherical object of the same weight .
In fig are iven the plan elevation of the path ained with a boonl e .
Ievatio upon a vertical designed to continue in its circular route as long as possible .
The arms of } ) meant are at right angles , and the twist and rounding a little ; the initial plane of rotation is vertical , and as much energy as possible is imparted in the act ) throwing , while the clim is slightly uphill .
The numerical value of is somewhat increased and thab of diminished , so that when the weapon in its return journey is over the thrower 's head , its axis of rotation , instead of being vertical , is inclined a little towards the inner side of the curve that it has described ; the forward velocity , though reduced , being still nnexpended , bhe original curve is continued , and the existence of implies that the plane of otatio will tilt slightly upwards and the tendency to fall be overcollue .
After the end of this second loop has been resched , the forward velocity has still further diminished , and gravity brings the boomerang , still spinning in a nearly horizontal plane , to the ground the starting-point .
I have obtained second loops , which were thirty yards in neasnred h , while , if the point be in , a lool ) will be described Plan .
tion aloug CAG .
In is represented the of a boomerang , of which the form an ngle which is larger about than that of the oviotS C Fig. 10 .
Fig. 11 .
Elevation on the ] through Plan .
Scale Fig. 1.3 .
Iovation on a plane parallel to DB .
MR. G. T. ON BOOMERANGS .
of rotation may point at the outset rather and the initial direction of motion is slightly uphill .
As the theoretical angular velocities indicate , there will be an increase in the value of , and this leads to the plane of rotation being horizontal when , he implement passes oyer the tbrower 's head at E. The angular velocity along the axis OX will then turn the path to the right along , while implies the body rises from to F. After a short time the forward velocity has diminished so that the descent from to the ground is made quite slowly , under the influence of gravity checked by the rotation of the boomerang in a nearly horizontal plane .
Figs. 10 , 11 illustrate the magnitude of the changes in the trajectory rendered possible by small variations in the shape of the missile .
This path was traced by a boomerang which subsequently warped to a slight extent in such a way as to increase the twisting : the natural flight then became the figure of eight of the two previous diagrams .
Through the kindness of Mr. O. ECKENSTEIN , I have recently had the opportunity of seeing and owing some boomerangs made by him , in which rounding present , but no twisting ; the allgle between the arms was considerably more obtuse , the size increased , and the weight doubled .
An examination of the equations ( 3-7 ) will show that if the value of be increased , the term due to gravity might be expected to replace for the most part that due to the twisting ; further , as the angle between the arms is larger , a given amount of will produce a greater effect in When the proportions are rightly chosen , I have not found it difficult to obtain a return path ; the plane of rotation is initially inclined at instead of to the horizon , amd with a decidedly smaller forward velocity as much spin as possible be imparted .
In the hands of one accustomed to its use , a boomerang of type is capable of extremely interesting flights .
For the remarkable diagrams ( figs. 12 , 13 ) which illustrate one of these , I am indebted to Mr. ECKENSTEIN .
VOL.

