]\gt ; Variation in Portunus depurator .
in air .
From the modification of their appearance , and the measurements originally made from them , their identity was not quite apparent , although probable .
It thus appears that , with the spark , the cyanogen spectrum is nothing like so strongly marked , as is the case with the flame of the gas , only one group of bands being represented , and that when the spectrum is taken in air the cyanogen does not appear , because in all probability the substance is oxidised .
I have already stated that the formation of cyanogen which yields the characteristic spectrum is a synthetical operation , that compound substances , when burnt in flames , do not , as a rule , emit the spectrum of the compound , but the spectrnm of one or more of the elements of which it is composed , or that of one or other of its products of combustion .
How then are we to account for the cyanogen spectrum in the flame of burning cyanogen conditions under which combustion takes place are these : there is an excess of the gas , the temperature of the is exceedingly high , and the gas within it is not in contact wibh a solid sub}Qtance , hence immediate decomposition doss not occur , and the gaseous compound heated to incandescence .
" " Variation in Portunus depurator By ERNEST WARREN , B.Sc. , trator of Zoology at University College , London .
b.y W. F. R. , F.B.S. Received Jnly 1 , The following measul.ements were undertaken at the proposal of Professor W. F. R. Weldon , and to him I am greatly indebted for many suggestions , and for the kindly help he has always so readily given me .
The were obtained fro1n the Biological Station at Plymouth , and sent at intervals during a period of aboul bwo years , dating the autumn of 1893 .
Only males were measured .
Seven measurements made on each individual , corresponding to those made by Weldon on the female of reinus moenas ( ' Roy .
Soc. Proc vol. 54 ) .
1 .
Carapace length , AB ( fig. 1 ) .
2 .
Total carapace breadth , 3 .
Frontal 4 .
Right antero-lateral , 5 .
Left antero-lateral , 6 .
Right dentary margin , 7 .
Left dentary margin , Mr. E. Warren .
The total number of crabs measured was 2300 .
The determinations were made with compasses and -decimetre ivory scale divided into half millimetres .
The measurements were recorded to the tentb of a millimetre .
As a test of accuracy , fifty crabs were indiscrimi .
nately taken out of a large number which had previously been measured ; these were remeasured , and the results compared with those before obtained .
It was found that the mean difference between any two measures of the same dimension was mm. FJG .
l.\mdash ; Portunus depurator 8 ( Plymouth , 1898-95 ) .
Continuous outlins represents a crab with carapace length AB mm. Dotted represents a crnb with carapwe length AB mm. Since the crabs were growing , and so varied much in size , it was necessary to reduce the measures to entages of some standard dimension .
The carapace length ( AB ) was selected , and the measurements are expressed as thousandths of this dimension .
The first point to ascertain is whether the mean of the different " " organs\ldquo ; varied the crab increased in size .
To determine this , the crabs were sorted into groups , according to the length of their standard dimension , then the arithmetic mean was found for the remaining six dimensions .
In each group the change in length of carapace was comparatively small , and had to be neglected .
The following table gives the results which were thus obtained : \mdash ; Variation in Portunus depurator .
Variation in Portunus depurator .
226 Mr. E. Warren .
no marked tendency either to rise or fall .
With these 1923 individuals a curve of frequency was drawn ( fig. 2 ) .
Its constants were calculated by the method employed by Professor Karl Pearson Phil. Trans vol. .
Range of variation thousandths of standard , unit of deviation is of carapace length ; therefore the observed range units , reckoning from 1227 upwards .
Centroid verbical position of arithmetic mean ) ] The second moment about the centroid Standard deviation ( error of mean square ) , Third moment Fourth moment , which is The critical function is positive , and so the theoretical curve has an unlimited range .
Professor Psarson 's measure of skewness for a curve of unlimited lange is given by the formula where Here ] and skewness .
It is clear from the values of the constants that the generalised probability curve wonld not differ perceptibly from the symmetrical normal curve , where and The areal deviation of the curve of observation from the normaI curve is only per cent. of the whole area .
Frontal Breadth\mdash ; It will be seen from the table that throughout the life of the crab the mean of this dimension falls steadily ; as the crab grows the forehead becomes relatively shorter .
On this account .
it is difficuIt to obtain a satisfactory idea of the distribution of deviations .
The means of groups 6\mdash ; 7 do not differ widely , and so wit , these the constants of variation wers calculated .
The observed range throughout the whole series was 640\mdash ; 795 thousandths of standard , and so there are 39 of our units of deviation .
The range in groups 6\mdash ; 7 ( including 460 crabs ) was 648\mdash ; 747 thousandths , that is , units .
Centroid 3.505490 . .
8kewness Heal again the critical function is positive Variation in Portunus depurator .
and so the generalised probability curve would be one of unlimited range .
R. Antero-lateral.\mdash ; The mean fluctuates considerably in the different groups , but thers is a slight tendsncy to fall .
omitting the first as being too small ] , it will be seen the mean falls about one unit in the whole series .
Range of deviation thousandths of standard , giving 25 units of variation .
We select groups 6\mdash ; 12 ( 1432 individuals ) for forming a frequency curve .
Here the range thousandths , giving 21 units .
FIG. .
Antero-lateral .
IndividuaIs fig 2 .
Centroid Skewness Variation in Portunus depurator .
FIG. .
Antero-lateral .
1432 Individuais ( see fig. 2 ) .
The critical function is positive , and so the range is unlimited , as was the case with the total and frontal breadths .
Using Professor Pearson 's skew of unlimited range , and we have where The distance of axis from centroid , and the distance of maximum ordinate from centroid .
Both this and the normal curve are drawn , and there is but very little difference between them , the areal deviations being and per ceut .
respectively .
R. Dentary Margin.\mdash ; The mean has a range of about units , and it rises as the crab grows .
The total range of deviation thousandths .
As in the other cases , oups 6 were selected , and a frequency urve drawn ; the range is 436\mdash ; 531 thousandths , giving 24 units .
Mr. E. Warren .
Csntroid .
Skewness The critical function is negative , and so the theoretical curve has a limited range .
This range is ; in the actnal statistics it is 24 , and so hers as in the case of the R. ateral it much exceeds any conceivable limit that may exist for the crab .
L. Dentary Margin.\mdash ; The trend and range of the mean resemble those of the R. dentary .
The total observed range of is thousandths .
In groups 6\mdash ; 12 the range is 425\mdash ; 524 , iving 2 units .
Centroid .
Skewness The critical function .
From this we see that the .
theoretical curve has a limited range ; but this range would be enormous , and the curve would closeJy resemble a normal curve .
Correlation of the Organs.\mdash ; Out of the organs discussed , the frequency curves of three of them ( total breadth , frontal breadth , and .
antero-lateral ) give theoretical curves of unlimited range , while the other three R. antero-lateral , R. and L. dentary margins ) give curves of limited range .
In every case the amount of skewness is small , and the diagrams show that the generalised probability cllrves do not give very obviously better fits than the normal curve .
The fact of the R. antero-lateral giving a strictly limited range while the L. antero.lateral gives an unlimited one , demonstrates that little stress can be placed upon the type of curve which a series of observations may yield .
In the present case the curves would appear to be.sufficiently normal to allow us to find Galton 's function ( known as r ) for pairs of organs .
For this purpose we shall employ the modified formula riation in Portunus depuratol where number of individuals , and and the standard deviations of organs A and respectively .
This formula has recently been shown by Professor Pearson to be superior to the one formerly used ( ' Phil. Trans vol. The correlation surfaces are published as being of permanent value , and it is believed that such material as this will be of use in the future in ating or modifying the CUl'rent theory of correlation .
Figs : 5 and 6 represent two lines .
The crosses were obtained by FIO . .
and L. Antero-lateral .
Dotted line the line inclined to axis .
The continuous line joining the crosses is line found on taking R. antero-lat .
as subject .
The circles indicate the points obtained lvhen the antero-lat .
is the subject .
The orobable errors of the dimensions are by the , \amp ; c. , on both horizontal and vertical scples .
Mr. E. Warren .
FIG. and B. Dentary .
Dotted line is the line inclined to axis at .
The continuous line joining the crosses is the line found on taking the B. antero.lat .
as subject .
The circles the points obtained when the R. the subject .
The probable erlors of the dimensions are represented by intervals , 1 , 2 , \amp ; c. , on both horizontal and vertical scales .
taking one dimension as the subject and the other as the relative , then for every unit of deviation of the snbject the mean of the associated values of the relative was found .
Ths values of the subject and the associated mean values of the relative are expressed in terms of their probable errors , and plotted along the axes of and respectiveIy .
The points indicated by circles were obtained by reversing the positions of subject and relative .
It is clear that the crosses and circles do tend to lie along a line Variation in Portunus depurator .
inclined at to the axis of .
The ends of the lines are irregular on account of the impossibility of obtaining a satisfactory[mean to the relative at these points because of the paucity of individuals near the limits of the range of deviation .
One secs that , these two cases at least , the correlation surfaces must closely approxlmate to the symmetrical normal surface .
In the above table the 2nd and 4th columns give the veins of Galton 's functions which ofessor Weldon found for two races of female Carcinus moenas ( ' Roy .
Soc. Proc vol. 54 ) .
The sixth column gives the values of obtained for Portunus .
It is quite obvio that there is a marked similariby between the three of figures .
The probable errors of were found by the formula , which Professor Pearson shows will give a close ap ( ' Phil. Trans vol. 187 ) .
These probable errors were added to indicate how far the differences in the values of are to be regarded as meaning actual deviations in the constants .
The values obtained for the two races of Carcinus diffel from one another nearly as much as they do from the constants of Portunus .
We have proved that the mutual reIationships of the organs measured are almost as closely similar between the two genera PortuMr .
E. Warren .
nus and Carcinus as between two not very sharply marked ofl of a single species .
Of course a considerable number of such comparisons would necessary before any safe conclusions could be drawn , and the mean .
ing of the differences observed could only be discovered by such a comparative treatment of a Iarge ssries of genera .
It is probable that larger deviations do indicate real differences in the correlation constant , possibly such are associated with changes in habit or environment .
For example , it is conceivable that a crab which swims might require to be more symmetrical than one that only crawls between the tide-marks .
Portunus does swim to a certain extent , and one can see from the table that the correlation of the two sides of the bodyis greater in this genus than in the essentially shoreliving Carcinus Variation in Portunus depnrator .
I. Correlation Surface of Total Breadth and R. antero-lateral .
1432 Individuals .
Mr. E. Warren . .
Correlation Snrface of Total Breadth and R. Dentary .
1432 Individuals .
Portunus IV .
Correlation Surface of Frontal Breadth and R. Antero-lateral . .
CorreIation Surface of Frontal Breadth and R. Dentary .
460 Individuals .
VI .
Correlation Surface of Frontal Breadth and L. Dentary .
460 Individuals .
VII .
Correlation Surface of R. Antero-lateral and L. -lateral .
1432 Individuals .
VIII .
Correlation Surface of R. ateral and R. Dsntary 1432 Individuals .
IX .
Correlation Surface of R. Antero-lateral and L. Dentar .
1432 Individuals .

