of construction of the residences) may yield useful results. 
Lack of information on population and housing is a particular handicap 
in the rapidly urbanising areas of developing countries. Even the results of 
decennial censuses are of only limited value in cases where a population growth 
of fifty per cent between censuses may be considered normal, and examples 
of over a hundred per cent not out of the question (Masser 1974). In such 
circumstances it is necessary to explore survey methods suitable for updating 
census material. There is also a need for short term forecasting methods based 
on the extrapolation of trends derived from sequential data on population and 
housing. The idea of using aerial photographs for this purpose is not new, but 
the yield of methodological publications in this field is extremely meager 
(Kraus et al 1974). Experts from census offices, and demographers, do not 
seem to consider it their task to develop a methodology for this. Geographers 
of the ecological school hardly use aerial photographs, so little can be expected 
from that quarter. Consultants from the industrial world, when faced with the 
lack of information, use aerial photographs without having the time to go 
deeper into the methodological aspects of their application (D'Alleux et al 
1975). However, to enable wider use of aerial photographs to occur in this 
field, it is necessary to increase research. The correlation between the classi- 
fications of residential areas based on artefacts and on socio-economic char- 
acteristics of the population, as well as the homogeneity within these classes, 
has been mentioned. Other questions will be indicated here briefly. When inter- 
preting aerial photographs, it is not always possible to distinguish between 
residential-like and residential-real, certainly not by reason of the mixture 
with home industry and retail activities. What is therefore the best operational 
definition of “residential”? And what, for a particular class of residential 
area is the most suitable characteristic for measuring the size of that area? Is 
it the housing unit that can be counted, or does measuring the area occupied 
by these houses with their inner courts and yards provide better results? What 
are the implications of using a crossgrid for area measurement (see figures 2 
and 3)? If interpretation and measurement are repeated by the same (or by 
another) interpreter, how different will the result be? What kinds of errors 
are made by the interpreters? 
Once the measurements have been made on the aerial photographs, 
these data have to be converted into population data. One should therefore,