112 A First Stiidi/ of Natm^al Selection in Clausilia laminata 
Mr Horace Darwin, and constructed by the Cambridge Scientific Instrument 
Company, which consists essentially of a long and flat piece of plate glass capable 
of movement in a plane, parallel to itself, under two compound microscopes. The 
part of the glass plate which moves under one microscope carries a scale, divided 
into tenths of a millimetre, while the microscope over it has a vernier in the eye- 
piece ; the part of the plate moving under the second microscope carries the object 
to be measured, the microscope over it having cross wires in the eye-piece. The 
object is adjusted so that the organ to be measured is parallel to the line of 
motion of the glass plate, and the scale is read when first one extremity of the 
organ measured, and then the other, lies under the intersection of the cross wires, 
the difference between the two readings of the scale giving the required length. 
The only serious source of error is a slight uncertainty which sometimes exists 
about the exact position of the apex ; but this does not, I believe, involve an uncer- 
tainty of 0 01 mm. in Mxy but the uppermost measures, which are unfortunately 
also the shortest. An inspection of Fig. 1 will show that a small uncertainty in 
the position of A (which can only affect its position in a direction perpendicular 
or neai'ly so to the axis of the shell) will not have an important effect upon 
the estimated length of any I'adius inclined at an angle of less than 45° to the 
columella. The error from obliquity in the plane of the sections measured is 
certainly too slight to affect measures involving only quantities of the order of 
001 mm. 
Before the measures of the radii of two shells can be compared, it is necessary 
to find some way of determining the plane in which the section of either shell cuts 
it. It is clear that an infinite number of planes can cut the shell so as to pass 
through the axis, and there is no obvious means of finding out which of these 
contains the actual section measured. There is therefore no means of knowing the 
angle through which the first radius measured in any section has revolved before 
meeting the plane of the section. Since the angular distance of the plane of the 
section from the origin of the spiral cannot be determined, it is necessary to choose 
some arbitrary plane to which each section may be referred ; and after consultation 
with Professor Pearson, all sections were referred to a plane containing a radius of 
the columellar spiral 5 mm. in length. 
The distance of such a plane from the plane of any actual section was found by 
interpolation between two adjacent columellar radii. For example in Fig. 1, if the 
columellar radius AG were exactly 5 mm. in length, the plane of the section would 
exactly coincide with the plane of reference ; if however AC were (as in the first 
individual of the series) -i'GH mm. long, and the next columellar radius Ad' were 
5'27 mm. long, then the plane of the section would be inclined to the plane of 
reference at a certain angle. Assuming the columellar spiral to be sensibly equi- 
angular through 180° (an assumption fairly in accord with observation) the angle 
between the plane of the section and the plane of reference may be taken to be 
