238 
MESSRS. R. H. FOWLER AND C. N. H. LOCK ON 
large change in the type of motion. This is shown clearly in fig. 3, round III. 16, 
where the decrease in amplitude and the change from an unstable to a stable type is 
quite marked, but still not sufficient to introduce any error in the determination of 
the couple for a single half-period. The effect is illustrated by the change in s in 
successive half-periods (Table III.), which is in general in the direction, and roughly of 
the amount, required by theory. 
Fig. 3. The observed and calculated motion in yaw, compared for selected rounds. The plotted 
points show the observed values of sin ifS plotted against Qt for rounds III. (11, 16) and IV. (8, 9). The 
continuous curves are the result of calculations described in detail in §§ 6, 8. Short vertical lines mark 
the positions of maxima and minima, and the origins of co-ordinates. The values of the constants used 
are as follows :— 
III. 11. (a) k = 80 degrees ; (b) k = 85 degrees, g 2 = 0 ; (c) k = 85 degrees, g~ — — 0-185. 
III. 16. (a) k = 80 degrees ; (b) k = 60 degrees, /? = 0 ; (c) k = 60 degrees, sin = 0-037 ; 
(d) k — 40 degrees, sin = 0-037 ; (e) k = 0, with third order contact with (a) at maximum. 
IV. 8. (a) k = 85 degrees ; (6) k = 75 degrees, sin = 0-034. 
IV. 9. (a) k = 60 degrees ; (b, c) k = 70 degrees. 
It appears that a change in M with v cannot alter an initial rosette motion into one 
with non-zero minimum yaw. This alteration, as in the stable case, must be due to 
the other couples depending on the angular velocity of the axis, and to the sideways 
