These operations require work on both faces alternately--this presents
no difficulties; but what appears to us most difficult to realize is
_continuous work_, the bar passing through several machines which
successively impress upon it the steps of progress toward the finished
chain. If the machines are end on to each other in a direct line, there
will necessarily be a fixed place for each tool; the rough cut chain
must accurately reach the point where another tool is ready to continue
the modeling. This appears to us practically impossible, the more so as
the elongation which the bar takes at each stamp varies with its initial
diameter.

What is more admissible is that with one heat and in the same machine an
operation could be performed on the two faces perpendicularly. The bar
could then be taken from one furnace and put in another immediately,
to pass at once to another machine to again undergo the operations
following. The work could then be done rapidly, submitting the bar to
several heats.

A few words on the tools as they exist.

The most important principle to note, and on which the different
machines employed are designed, is this: The punches or matrices acting
on the chain at its different points of progress are put in motion by
spiral springs worked by means of tappets or cams distributed over the
circumference of a cylinder, having a rotary movement imparted to it by
pulleys and belts.

The figures on our plate show with sufficient clearness the working of
one of these machines. It will be seen that the bar traverses through