August i, 1901] 



NA TURE 



539 



constant length measured along it. The scale of these diagrams 

 is about 1: 1000). 



The angular velocity (l) is increased by an increase of twist 

 and bv an increase of rounding ; it also increases when cos 6 

 increases, where 6 is the inclination of the plane of rotation to 

 the horizontal. The curling to the left (2) is increased by an 

 increase of twist, or of cos », and, in general, by an increase of 

 rounding. 



Fig. 5.— Ele 



1 through CA. 



If it be desired that the boomerang should describe a second 

 circle in front of the thrower (Figs. 6, 7), it must be thrown 

 much harder, so that when one circle has been described it may 

 still have sufficient forward velocity. When the projectile has 

 described the first circle and is over the thrower's head, the axis 

 of rotation must point in an upward direction in front of him ; 

 if it pointed behind him the subsequent path would be behind 

 his back, and a figure of eight (Figs. 8, 9) would become pos- 

 sible. For a path with a second loop in front of the thrower 



Fic. 7. —Ele 



through c E. 



he should acco rdingly choose a boomerang with much twist and 

 much rounding , and throw it with his body leaning over to the 

 left, so that th e angle S between the axis of rotation and the 

 vertical may be slightly in excess of a right angle. The increased 

 twist will mean that the first circle has a smaller circumference 

 and that there will be more pace left after it has been described ; 

 and the increased rounding will keep the plane of rotation from 

 becoming hori zontal too soon. 



For a figure of eight we should require less rounding, or we 



NO. 1657, VOL. 64] 



might give more spin in throwing, and aim a little uphill with 

 e rather less than a right angle. There are so many elements 

 capable of variation that nothing but experience can teach how 

 to get the best results with any particular boomerang. 



The most complex path that the author has succeeded in 



effecting is that of Figs. 10 and II. But it is certain that these 

 fall far short of what is done by skilful natives of Australia. 



If the angle between the arms is increased and the twist and 

 rounding unaltered, the angular velocity (l) is mcreased, and it 

 becomes easier to make a second loop behind than in front. If 

 the angle exceeds 150', the angular velocity of the first kind is 

 so large that it is very hard to get a return at all. 



Fig. io.— Plan. 



When the twist is left-handed and the angle large we have a 

 specimen of the second type (Fig. 3\ and it must be thrown 

 with the more rounded side uppermost and the plane of 

 rotation inclined at between 30" and 60" to the horizontal 

 (i.e. 30° < e < 60°) ; the angle of projection (i.e. inclination to 



