1034 
PROFESSOR J. W. MALLET OR A REVISION 
Of course one distinct exception to the assumed law would disprove it, if that 
exception were itself fully proved, but this is not the case. 
As suggested by Marignac and Dumas, anyone who will impartially look at the 
tacts can hardly escape the feeling that there must be some reason for the frequent 
recurrence of atomic weights differing by so little from accordance with the numbers 
required by the supposed law. 
As the question stands at present, the following 18 atomic weights are the only ones 
which may be fairly considered as determined with the greatest attainable precision, 
or a very near approach thereto, and without dispute as to the methods employed— 
“Oxygen 
^Nitrogen . 
Chlorine . 
Bromine 
Iodine . 
* Sulphur 
'“Potassium . 
'"Sodium 
'"Lithium 
Silver . 
Thallium . 
'"Aluminum. 
'“Carbon . 
'"'Phosphorus 
Barium . 
Calcium 
'"'Magnesium 
Lead 
15-961 
14-010 
35-37 It 
79757+ 
126-541 + 
3L996+ 
39-042 
22- 987 
7-005 
107-667+ 
203-655} 
27-019 
11-97 
30-96 
136-84 
39-90 
23- 94 
206-40 
If now we discard altogether Dumas’ assumption of multiples of ’5 or *25, and 
consider simply the indications afforded of Prout’s law in its original form, we may 
safely take the first decimal place of each of these numbers as quite freed from the 
influence of fortuitous errors, while the second decimal is nearly so in many instances. 
It appears that out of the 18 numbers, 10 (those to which an asterisk is prefixed) 
approximate to integers within a range of variation less than one-tenth of a unit. 
What then is the degree of probability that this is purely accidental, as those hold 
who carry to the extreme the conclusions of Berzelius and Stas ? Since there are 
t Stas’ numbers, uncorrected for occlusion of oxygen by silver, 
t Crookes’ number modified by taking 0 = 15'961 instead of 15'96. 
