96
J . M . LAMBERT AND M . B . DALE
serious consideration ; moreover, there is quite rightly considerable
suspicion of many of the more empirical methods which have been
suggested at one time or another for the solution of particular problems;
and again, as Greig-Smith points out (1964, p. 211), there is always the
danger of such excessive preoccupation with techniques that the
problems themselves may be pushed into the background. Such considerations have probably been at least partially responsible for the
rather sceptical attitude of some ecologists as to the general usefulness
of a statistical approach. It is also perhaps inevitable that the theoretical advantages may frequently seem to be outweighed by sheer production difficulties : many of the potentially valuable techniques are
still under development, progress is often held up through lack of
necessary fundamental mathematical research, and few computerprogrammes specifically designed for ecological work have yet passed
the stage of the small-scale prototype model.
Most working ecologists are naturally more concerned with the
question of practicability than anything else : however elegant or
powerful a method, it is of little interest to them unless it can actually
be used. Unfortunately, this has sometimes led to the use of relatively
simple but mathematically unsound techniques ; and, although the use
of such methods can never be profitable, a sound but too-elaborate
method may defeat its own purpose by never being adopted. The
essential need in ecology, therefore, is for a series of well-founded but
relatively crude techniques in which the degree of approximation is
clearly understood and recognized.
Even given the existence of such techniques, however, it is the scale
of most ecological data which is frequently regarded as the major
obstacle to their extensive use. But here there seems to be a gross underestimation by most ecologists as to the scale of problem it will be
possible to tackle with modern computer facilities once the programmes
have been made. Advances in computer design have recently been so
great that statistical analysis of vegetation on a regional scale can now
be contemplated as a practicable possibility within the next few years.
For instance, the possibility of a large-scale computer programme for
association-analysis, capable of dealing with the whole of the B.S.B.I.
Distribution Maps Scheme data of some 2 000 species in some 3 500
10-km squares (Walters, 1954), is now under active consideration;
while a programme already exists for the Elliott 803 computer which has
performed both normal and inverse analyses for 317 species in 152 vicecomital sites in about 40 h for the two analyses together (M. C. F.
Proctor, in Zitt., 1964).
Although the time taken to run a.particular computer-analysis is
partly a function of the scale of the data, it is also very dependent on
J . M . LAMBERT AND M . B . DALE
serious consideration ; moreover, there is quite rightly considerable
suspicion of many of the more empirical methods which have been
suggested at one time or another for the solution of particular problems;
and again, as Greig-Smith points out (1964, p. 211), there is always the
danger of such excessive preoccupation with techniques that the
problems themselves may be pushed into the background. Such considerations have probably been at least partially responsible for the
rather sceptical attitude of some ecologists as to the general usefulness
of a statistical approach. It is also perhaps inevitable that the theoretical advantages may frequently seem to be outweighed by sheer production difficulties : many of the potentially valuable techniques are
still under development, progress is often held up through lack of
necessary fundamental mathematical research, and few computerprogrammes specifically designed for ecological work have yet passed
the stage of the small-scale prototype model.
Most working ecologists are naturally more concerned with the
question of practicability than anything else : however elegant or
powerful a method, it is of little interest to them unless it can actually
be used. Unfortunately, this has sometimes led to the use of relatively
simple but mathematically unsound techniques ; and, although the use
of such methods can never be profitable, a sound but too-elaborate
method may defeat its own purpose by never being adopted. The
essential need in ecology, therefore, is for a series of well-founded but
relatively crude techniques in which the degree of approximation is
clearly understood and recognized.
Even given the existence of such techniques, however, it is the scale
of most ecological data which is frequently regarded as the major
obstacle to their extensive use. But here there seems to be a gross underestimation by most ecologists as to the scale of problem it will be
possible to tackle with modern computer facilities once the programmes
have been made. Advances in computer design have recently been so
great that statistical analysis of vegetation on a regional scale can now
be contemplated as a practicable possibility within the next few years.
For instance, the possibility of a large-scale computer programme for
association-analysis, capable of dealing with the whole of the B.S.B.I.
Distribution Maps Scheme data of some 2 000 species in some 3 500
10-km squares (Walters, 1954), is now under active consideration;
while a programme already exists for the Elliott 803 computer which has
performed both normal and inverse analyses for 317 species in 152 vicecomital sites in about 40 h for the two analyses together (M. C. F.
Proctor, in Zitt., 1964).
Although the time taken to run a.particular computer-analysis is
partly a function of the scale of the data, it is also very dependent on