xxix.] EXCEPTIONAL PHENOMENA. 665 



that ill substances yet examined fall into one of two classes ; 

 they are all either ferro-magnetic, that is, magnetic in the 

 same way as iron, or they are diamagnetic like bismuth. 

 But it does not follow that every substance must be ferro- 

 magnetic or diamagnetic. The magnetic properties are 

 shown by Sir W. Thomson 1 to depend upon the specific 

 inductive capacities of the substance in three rectangular 

 directions. If these inductive capacities are all positive, we 

 have a ferro-magnetic substance ; if negative, a diamagnetic 

 substance ; but if the specific inductive capacity were 

 positive in one direction and negative in the others, we 

 should have an exception to previous experience, and 

 could not place the substance under either of the present 

 recognised classes. 



So many gases have been reduced to the liquid state, and 

 so many solids fused, that scientific men rather hastily 

 adopted the generalisation that all substances could exist 

 in all three states. A certain number of gases, such as 

 oxygen, hydrogen, and nitrogen, have resisted all efforts to 

 liquefy them, and it now seems probable from the experi- 

 ments of Dr. Andrews that they are limiting exceptions. 

 He finds that above 31 C. carbonic acid cannot be liquefied 

 by any pressure he could apply, whereas below this tem- 

 perature liquefaction is always possible. By analogy it 

 becomes probable that even hydrogen might be liquefied if 

 cooled to a very low temperature. We must modify our 

 previous views, and either assert that below a certain critical 

 temperature every gas may be liquefied, or else we must 

 assume that a highly condensed gas is, when above the 

 critical temperature, undistinguishable from a liquid. At 

 the same time we have an explanation of a remarkable 

 exception presented by liquid carbonic acid to the general 

 rule that gases expand more by heat than liquids. Liquid 

 carbonic acid was found by Thilorier in 1835 to expand 

 more than four times as much as air ; but by the light of 

 Andrews' experiments we learn to regard the liquid as 

 rather a highly condensed gas than an ordinary liquid, and 

 it is actually possible to reduce the gas to the apparently 

 liquid condition without any abrupt condensation. 2 



1 Philosophical Magazine, 4tL Series, vol. L p. 182, 



2 Maxwell, Theory of Heat, p. 13}. 



