5 sets of 50 observations for each measuring mark examined. In the 
comparison series, each point was based on the mean of 10 sets of 
10 readings. 
All settings were carried out with the observer's right eye 
without spectacle correction, with the head vertical throughout. 
The right eye was 1 Dioptre hypermetropic, and the left eye 3/4 
Dioptre hypermetropic. There was 3^Esophoria at 6m, and orthophoria 
at 33cm. No astigmatism was present. 
The time taken for a typical setting and recording was 1 minute 
or less. 
3. The Results 
The computed values of the pointing errors for all the cases 
investigated are located in the various tables in the Appendix, and 
are plotted in figures 2 and 3. The pointing error has been defined 
as Sp = 7s|T"sf, where S2 and S^. are the variances of a pointing 
derived from the pointings in the x and y directions respectively. 
The standard deviations of the means were computed on the basis 
of the assumption that the means are correlation-free. 
To facilitate comparison, the results of the earlier stereo 
comparator observations (10) are also plotted in figures 2 and 3. 
Inspection of the figures 2 and 3 and the appropriate tables 
reveals the following general tendancies. 
(i) The precision of the stereocomparator as used in the 
earlier investigations was insufficient to develop 
the full capacity of the eye, and masked completely 
the phenomena revealed in these tests. 
(ii) The relationship between pointing error and target 
subtense angle was complex within the subtense range 
investigated (2-45 minutes of arc). The nature of 
the task changed at a critical point and the form of 
the relationship was the same for measuring marks 
2-18 minutes of arc subtense. 
(iii) The critical variable was the width of the annulus 
between mark and target. The critical visual infor 
mation for the pointing appeared to be contained in 
an annulus 1 minute of arc wide between mark and target, 
and Weber's law applied to the relationship in this 
range. For annuli of width 1-4 minutes of arc, the 
pointing error was constant, and then there was a 
4