
’ 
ON THE ACTION OF AIR AND WATER UPON IRON. 5 
cally manifest from the fact before adverted to (2nd. Rep. 179), that the very 
same sort of iron corrodes much faster when cooled irregularly and fast than 
it does when the contrary has been the case. Of this we have instances in 
the irons a 8 and 9, @ 14 and 15, &c., of which analyses are given. Minute 
variations in the foreign alloying metals usually found in cast iron do not 
appear to effect its corrodibility, and the slight and uncertain difference which 
exists between hot and cold blast iron as to corrosion arises rather from their 
difference in specific gravity than anything else. 
It is observable also that the important improvement of the hot blast has 
in this respect little deteriorated the quality of cast iron, as our experiments 
(a 26, 27) show that iron made thirty-five years ago in Scotland before its 
introduction differs very slightly in corrodibility from that of recent manu- 
facture by hot blast. 
293. It will be remembered that carbon exists in cast iron in two very 
different states, viz. as diffused graphite in a crystalline form and as combined 
carbon ; that the dark gray and softer irons contain more of the former, the 
brighter and harder irons more of the latter. Now the latter kind have the 
property of being much less uniform or homogeneous of surface when cast . 
under similar conditions than the former, while the highly graphitic irons, 
though more uniform in large specimens, are the least dense and softest in 
texture: hence the ultimate choice at which we arrive is, that the bright gray 
irons of high commercial marks, the No. 1 and 2, while they are in all other 
respects the most valuable for construction, are also the most durable. 
294. Voltaic uniformity of surface is best attained by slow cooling of the 
metal when cast, and in all small castings will be much promoted by subse- 
quent annealing out of contact of air, as in the process ordinarily used for 
decarbonizing cast iron to render it flexible and tough. 
295. As the analysis of cast iron is admittedly a matter of some difficulty, 
to ensure trustworthy results, it may be proper to state briefly the methods 
pursued with those above given and with some others which it was needless 
here to bring forward. 
One of the principal difficulties exists in the determination of the carbon; 
for this a number of methods have been proposed. Berzelius burnt the 
carbon by passing a slow current of dry oxygen over the pulverized metal, 
absorbing the carbonic acid by barytic water. He also proposed a similar 
process with dry chlorine, volatilizing the chloride of iron formed; and the 
methods by ‘chloride of silver or copper. 
Berthier devised a process by dissolving the metal in iodine or bromine, 
the object held in view by all being to avoid the loss of carbon which inevi- 
‘tably results from solution of the’ metal in acids evolving hydrogen. All 
these modes however are so tedious and beset with practical difficulties as to 
give uncertain results. 
The method adopted by me in most cases was a modification of Regnault’s 
process, which consisted in mixing the cast iron finely pulverjzed with about © 
twelve times its weight of chromate of lead properly prepared and mixed 
with a little chlorate of potass. This is burnt in an ordinary combustion-tube, 
in the remote extremity of which some dry powdered chlorate of potass is 
placed, and heated after the combustion has been completed, so as to pass a 
current of oxygen over the ignited mass. This precaution is indispensable 
with the harder and denser irons containing most of their carbon in combi- 
nation. The total amount of constituent carbon is thus obtained and weighed 
as carbonic acid ; but this consists of graphite and of combined carbon. By 
a separate assay the graphite is obtained by solution of a weighed portion of 
the metal in nitric acid, as residue consisting of graphite, extractive matter 
