484
The influence of tobacco smoking upon operator’s precision with
photogrammetric instruments
Introduction
Tobacco smoke contains nicotine, tar, carbon and other substances
which are inhaled by the smoker. This is an investigation of the possible
influence of smoking upon stereoscopic vision.
Experimental Design
Stereoscopic models were formed from aerial photographs placed in
the Wild and the early type of Zeiss stereocomparator. Clearly identi
fiable points in the models were selected for measurement. The paral
lax in each point was measured stereoscopically with 25 measurements
forming a series. Several series were measured before and after heavy
smoking. A mean value (px, py) and a standard deviation (s px , s py )
were computed from every series. The data were then compared to
decide the influence of smoking. Three operators took part in the tests
during a period of two days. Assistants read and recorded their mea
surements. Operator No 1 measured the horizontal parallaxes of two
different points in the Zeiss comparator. Operators Nos 2 and 3 mea
sured the horizontal and vertical parallaxes of the same point stereo
scopically in the Wild comparator.
Statistical Treatment of Data
Differences between the standard deviations in the series measured
by the same operator were analyzed by Bartlett’s test. In most cases
this test shows a wide variation in the precision. Therefore we cannot
use variance analysis to investigate the differences of the mean values
as it is impossible to state whether the differences depend upon the
standard deviations or the mean values or both. However, a difference
between two mean values can in any case be tested with a t-test even
if the standard deviations differ. The difference between two standard
deviations is tested with an F-test. (See reference No 1 for more details
on statistical procedure.)
Residts of Tests
The two points measured in the Zeiss comparator were not equally
clear in identification. This was evident from the standard deviations.
One day the precision was uniform, the next day it varied. In both
cases an analysis showed differences between mean values which were
higly significant. The influence of smoking could not be isolated.
In the Wild comparator px and py were measured in the same series.
The differences between s px and s py in the same series were not signi
ficant for any one operator. A study of the variation of s px and s py
during the course of a clay indicated a uniform s py for operator No 2