Between these two census years the annual end-of-year population for each of the dif-
ferent territories in the area has been estimated by various official sources as in Table 2a.

 Comparing on an overall basis the figures in Table 2a with those given in Table 2. 1. i,
we notice that the 1960 census figure of 3,669,528 is less than each of the end-of-year
estimates since 1957. The discrepancy between the 1957 end-of-year estimates and the
1960 census is 14,859 and by 1959 it rises to 93,942. There are several reasons for this,
but perhaps the most important is the method of computing the annual end-of-year popula-
tion estimates.

 Briefly, they have been derived by taking the previous year's population estimate for a
particular territory and adding to it the net increase due to the excess of births over
deaths, and deducting or adding, as the case may be, net migration for the particular
year. So the end-of-year population for any territory in 1950, say, would be computed
as follows:

 End-of-yr. pop. (1950) = End-of-yr. pop. (1949) + Net Incr. 1950  Net Migrn. 1950.

The individual estimates for each territory are then aggregated to derive the overall total
for the area for that year.

 This method of computing the end-of-year estimates suffers from one very obvious and
basic defect. If the original base of the estimate is wrong or overestimated, as is the
case for some of the territories, then the end-of-year estimates will be out. Naturally,
the error becomes cumulative as we derive end-of-year estimates for any considerable
period of time. It would appear, therefore, from the figures given in Table 2a that most
of the territories have been consistently overestimating their end-of-year population.
So if we are to avoid making the same mistake of using inflated base estimates for future
years, it is necessary to strike out with some new estimates which start from a proper
base. It should be noted, however, that population figures quoted for years previous to
1960 in any part of the text, are the official year end figures.

 Data at present available from the 1960 census allow us to attempt only the simplest
projections. We offer two projections which are based mainly on information from the
1946 census and such information from the 1960 census as is available.


Projection 1
 The assumptions underlying this projection are quite simple and unsophisticated. From
the figures given in Table 2. 1. i we know that the population of the area increased from
2,776,056 in 1946 to 3,669,528 in 1960, an intercensal period of 14 years. If we assume
that the census population of 2,776,056 in 1946 increased by some constant rate of growth
which, when compounded on an annual basis, would give us the census population of
3,669,528 for 1960, then we can derive a natural exponential function of the form y = aetr,
where a is a constant, e = 2.71828, the base of the natural log, t = 14, the number of
intercensal years and r = the annual compounded rate of growth.