Karl Pearson and David Heron 
285 
Forza e Ricchezza, Turin, 1906, p. 15. For the rents we have not been able to 
verify Professor Niceforo's returns, or to discover whether they are for approxi- 
mately the same years as the statures. In the Annuaire Statistique de la Ville cle 
Paris, 1901, no similar details as to rent appear to be given, and Professor Niceforo 
gives no reference to the year for his rent data. Taking, however, the data given 
in his book we find : 
Mean Stature =1645-75. Mean Rent* =210-8. 
Standard Deviation = 6"212. Standard Deviation = 161 '43. 
Correlation of Stature and Rent = "7825 + "0591. 
Regression line of Stature 8 on Rent is!: 
S= 1639-4 + 03011 R. 
This is represented on Diagram XXIV and we see at once that we have a quite 
intelligible relation between average rent in the arrondissement and average 
stature of conscripts. The correlation is high, but far from " perfect," and is subject 
Diagram XXIV. Association of Stature and Rent in Parisian Arrondissements. 
1630 ] , , 1 , , , , , 
50 100 150 200 250 300 350 400 450 500 550 600 650 700 750 800 
Mean Rent in Parisian Arrondissement. 
to a large probable error, so that the true relation might be easily anything between 
•66 and "90. If, however, we divide at the mean values and add up all the dots as 
1 I 8 
Professor Niceforo appears to have done, we have the fourfold table: — - — — 
giving association Q = 1 + -0. What is the value of this, what does it signify ? 
We fail to extract the least idea of the real relationship — as represented by the 
dots on the graph — from such a statement. Verbally it means simply that no 
rent of over 210'8 is associated with a stature under 1645'75, but to assert that 
* We are unable to say in wliat units or for what periods rent is measured. 
