Thompson, Cells in the Cerebral Cottex of Man. 129 



Table IV, in which this calculation is recorded, shows that 

 the total number of functional nerve cells in the cerebral cortex 

 is 9,282,826,403. 



I. The chief sources of error in the preceeding calcula- 

 tion are two in number ; first, the additions to Hammarberg's 

 data ; and, second, the more or less arbitrary proportion of the 

 total area assigned to each lobe and to the regions within each 

 lobe. Reference to Tables VII and VIII, where additions are 

 indicated by parentheses, will show that the greater part of 

 them concern the first layer. The thickness of this layer, where 

 it is given, varies only from o. i to o. 3 mm. In most places it 

 is o. 2 mm. Wherever the thickness of the layer is not men- 

 tioned, therefore, o. 2 mm. has been adopted. The frequency 

 of the cells in this layer is nowhere mentioned. Hammarberg 

 says merely that they are very few in number. A uniform fre- 

 quency of five per unit cube was assumed. Now if this entire 

 layer were omitted from the calculation, it would change the re- 

 sult only about 2.5%. The absolute error involved in assum- 

 ing values which may not be perfectly accurate, is of course 

 much less than this. The other additions to the tables are 

 comparatively few in number, and cannot be very inaccurate, 

 since they are based on the values for closely similar regions, or 

 else are indicated by Hammarberg's verbal comments. It seems 

 safe to say, therefore, that the final result cannot be materially 

 affected by this source of error. 



The case in which a mistake in the correct proportioning 

 of the lobes would produce a maximum error is that of the 

 occipital lobe. The averages for the others differ so little, that 

 making one or another of them too large or too small could 

 not affect the final outcome appreciably. The unit column for 

 the temporal and parietal regions is in round numbers, 500, and 

 that for the occipital region, 750 (Table II). If the error lay 

 in the direction of making the occipital lobe too small, our re- 

 sult would be a little less than it should be, and the discrepancy 

 between this calculation and others previously published, would 

 only be decreased. But if the occipital lobe has been made too 

 large the final estimate is somewhat too large. The occipital 



