When the curve is all in deep water, and there is a pool after _B_, the boat ought not to be turned too quickly in endeavouring to avoid the rock _C_, else it will sometimes then enter the eddy below _B_, which runs up stream sometimes for fifty yards. In such a case the absurd position you are thereby thrown into naturally causes you to struggle to resist or stem this current; but I have found, after repeated trials of every plan I could think of, that if once the back current has taken the canoe it is best to let the boat swing with the eddy so as to make an entire circuit, until the bow can come back towards _B_ (and below it), when the nose of the boat may be again thrust into the main stream, which will now turn the boat round again to its proper course. Much time and labour may be spent uselessly in a wrong and obstinate contest with an eddy. In fig. 6, where the three rocks are in a straight line, and the middle one is a breaker, an instance is given when the proper course must be kept by _backing_ during the first part of it. We must suppose for this that the canoist has attained the power of backing with perfect ease, for it will be quite necessary if he intends to take his boat safely through several hundred combinations of sunk rocks and breakers. Presuming this, the case in fig. 6 will be easy enough, though a little reflection will show that it might be very difficult, or almost impossible, if the canoist could give only a forward motion to the boat. To pass most artistically, then, through the group of rocks in fig. 6 the stern should be turned towards _A_, as shown in the diagram, and the passage across the current, between _A_ and _B_, is to be effected solely by backing water (and chiefly in this case with the left hand) until the furthest point of the right of the curve is reached, with the boat's length still as before in the position represented in the figure. Then the forward action of both hands will take the canoe speedily through the passage between _B_ and _C_. Cases of this sort are rendered more difficult by the distance of _C_ from the point above _A_, where you are situated when the decision has to be made (and in three instants of time) as to what must be done; also, it would usually be imprudent to rise in the boat in such a place to survey the rock _C_ from a better position. If it is evident that the plan described above will not be applicable, because other and future circumstances will require the boat's bow to emerge in the opposite direction (pointing to the right), then you must enter forwards, and must back between _B_ and _C_, so as to be ready, after passing _C_, to drive forward, and to the right. It is plain that this is very much more difficult than the former case, for your backing now has to be done against the full stream from the breaker _B_.