7. The 'biceps' should not do the work; for, if it does, either the hands are elevated or the level of the blade altered--if the elbows keep close to the side; or else, if the level of the hands is preserved, then the elbows dog's-ear outwards. In either case the action is less free and less powerful than if the stroke is rowed home by the shoulder muscles. 8. The part of the hand which should touch the chest when the oar comes home is the root of the thumb, not the knuckles of the fingers. If the knuckles touch the chest _before_ the oar comes out of water, the blade is 'feathered under water'--a common fault, and a very insidious one. If, on the other hand, the oar comes out clean, but the first thing which touches the chest is the knuckle, then the last part of the stroke will have been rowed in _air_, and not in _the water_. 9. Dealing now with recovery. The hands should rebound from the chest like a billiard-ball from a cushion. If the hands delay at the chest they hamper the recovery of the body--e.g. let any man try to push a weight away from him with his hands and body combined. He will find that, if he pushes with straight arms, he is better able to apply the weight of his body to the forward push than if he keeps his arms bent. Having shot his hands away, and having straightened his arms as quickly as he reasonably can, his body should follow; but his body should not meantime have been stationary. It should, like a pendulum, begin to swing for the return so soon as the stroke is over. If hands 'hang,' the body tends to hang, as above shown; and if the body hangs, valuable time is lost, which can never be regained. As an illustration: suppose a man is rowing forty strokes in a minute, and that his body hangs the tenth of a second when it is back after each stroke, then at the end of a minute's rowing he will have sat still for four whole seconds! An oarsman who has no hang in his recovery can thus row a fast stroke with less exertion to himself than one who hangs. The latter, having wasted time between stroke and recovery, has to swing forward all the faster, when once he begins to recover, in order to perform the same number of strokes in the same time as he who does not hang. Now, although there is a greater effort required to row the blade square through the water than to recover it edgewise through the air, yet the latter has to be performed with muscles so much weaker for the task set to them that relatively they tire sooner under their lighter work than do the muscles which are in use for rowing the blade through the water. When an oarsman becomes 'pumped,' he feels the task of recovery even more severe than that of rowing the stroke. Hence we see the importance of economising as far as possible the labour of those muscles which are employed on the recovery, and of not adding to their toil by waste of time which entails a subsequent extra exertion in order to regain lost ground and lost time.