112 DIRECT PROPERTIES OF MOLECULES 305 



molecule, and may at the same time rotate about their own 

 axes. 



But in one point the similarity between a molecular and 

 a planetary system need not exist. In a planetary system 

 the plane of the motions is unalteringly fixed in space, or 

 else moves so slowly that the change is recognisable only 

 after a long time. But in a molecular system of atoms the 

 plane of motion may be looked upon as variable ; indeed it 

 alters its direction at every collision of one molecule with 

 another. 



The agreement of the tabulated figures seems to support 

 such a supposition of the condition of the molecules of all 

 gases which are diatomic only. To a pair of atoms of this 

 kind corresponds a double star in the stellar heavens, each 

 part of which describes a plane path about the common 

 centroid. 



Further, when three atoms are contained in a molecule 

 we must suppose the molecule to have a plane shape. 

 According to chemists' views the atoms are then so bound 

 together as to be arranged either in a straight line or in 

 a ring. In the last case the plane will be determined by 

 the three points at which the three atoms are. The motion 

 then takes place either by two atoms revolving round the 

 third, or by all three, forming a triangle, revolving about 

 their common centroid. The admissibility of these con- 

 ceptions arises from the agreement of the calculated with 

 the observed values of Q in all the cases that have been 

 cited, with the single exception of water- vapour ; the devia- 

 tion in this case, however, probably depends only on the 

 observed value not being reducible to the temperature C., 

 to which the other numbers in the table are referred. 



When four atoms are joined together to form a molecule 

 it is in general no longer necessary for them to possess the 

 property of being a plane system ; the possibility, however, 

 of the system being of such character is shown by the 

 example of ammonia, for which the calculated value of Q 

 agrees very exactly with that observed. We shall conse- 

 quently be unable to make any other supposition as to the 

 molecular constitution of ammonia than that usual with 



