This change in the action of the steam upon the piston rendered it necessary to make a corresponding change in the mechanism by which the piston-rod was connected with the beam. In the single acting engine, the piston-rod pulled the end of the beam down during the descent, and was pulled up by it in the ascent. The connection by which this action was transmitted between the beam and piston was, as we have seen, a flexible chain passing from the end of the piston and playing upon the arch-head of the beam. Now, where the mechanical action to be transmitted is a _pull_, and not a _push_, a flexible chain, or cord, or strap, is always sufficient; but if a _push_ or _thrust_ is required to be transmitted, then the flexibility of the medium of mechanical communication afforded by a chain, renders it inapplicable. In the double-acting engine, during the descent, the piston-rod still pulls the beam down, and so far a chain connecting the piston-rod with the beam would be sufficient to transmit the action of the one to the other; but in the ascent the beam no longer pulls up the piston-rod, but is pushed up by it. A chain from the piston-rod to the arch-head, as described in the single acting engine, would fail to transmit this force. If such a chain were used with the double engine, where there is no counter weight on the opposite end of the beam, the consequence would be, that in the ascent of the piston the chain would slacken, and the beam would still remain depressed. It is therefore necessary that some other mechanical connection be contrived between the piston-rod and the beam, of such a nature that in the _descent_ the piston-rod may _pull_ the beam down, and may _push_ it up in the _ascent_.