-------------------------------------+ | | | | Observed. | Calculated. | |------------------|-----------------| | Mean of the | | | corresponding | | | ordinates in | | | the six curves | | | after reduction | | | to the common | Ordinates to | | scale of | the normal curve| | p.e. = 100. | of distribution,| | 965 observations | probable error | | in all. | = 100. | |------------------|-----------------| | -231 | -244 | | -182 | -190 | | -117 | -125 | | - 93 | -100 | | - 73 | - 78 | | - 37 | - 38 | | + 1 | 0 | | + 38 | + 38 | | + 77 | + 78 | | +107 | +100 | | +139 | +125 | | +213 | +190 | | +260 | +244 | -------------------------------------+ Table XXXII. is derived from Table XXXI. by a process described by myself in many publications, more especially in _Natural Inheritance_, and will now be assumed as understood. Each of the six pairs of columns contain, side by side, the Observed and Calculated values of one of the six series, the data on which the calculations were made being also entered at the top. The calculated figures agree with the observed ones very respectably throughout, as can be judged even by those who are ignorant of the principles of the method. Let us take the value that 10 per cent of each of the six series falls short of, and 90 per cent exceed; they are entered in the line opposite 10; we find for the six pairs successively, _Obs._: 5.5 4.8 0.64 0.59 0.50 0.74 _Calc._: 6.0 4.2 0.67 0.51 0.48 0.68 The correspondence between the more mediocre cases is much closer than these, and very much closer than between the extreme cases given in the table, namely, the values that 5 per cent fall short of, and 95 exceed. These are of course less regular, the observed instances being very few; but even here the observations are found to agree respectably well with the proportions given by calculation, which is necessarily based upon the supposition of an infinite number of cases having been included in the series.