Mike is correct, the history of this goes back a long way.
One key document explaining the current philosophy was the post-war building studies number 29-fire grading of buildings.
The following text was copied from this document.
"The question of traffic movement in exits has been the subject of investigation and test at various times in both this country and abroad. Only a limited number of tests have been made in this country. Figures for eight tests have been published in appendices to the factories act 1937. More extensive tests have been made in America notably by the United States bureau of standards, the results of which are presented in their publication Design and Construction of Building Exits (1935). These tests have, with one or two exceptions, being carried out under random flow conditions i.e. in the course of the normal use of the buildings where the tests were made or during fire drills. The observations were of course made whilst the exits were being used to capacity as far as could be judged. It is useful to compare these tests with the results of experiments carried out by the Paris fire brigade with Firemen under controlled conditions. These results, published in 1938 and 1945, helped to isolate the effect of certain factors which have an influence on the rate of movement in exits."
There is then a section talking about the use of unit widths which for the purpose of their studies were taken as 21 inches.
To continue quoting from the document
"The results given in appendix 1 showing disconcertingly wide variation. The rate of movement on stairs varies between 19 and 107 persons per unit weight per minute the extreme figures can be readily discounted for practical purposes but there still remains a considerable variation to be taken into account in fixing a reasonable figure for determining means of escape requirements. Consideration of the test results on the available explanatory notes indicates that the features of traffic movement phenomena in which variation can occur may be considered under four headings:
urgency motive controlling speed of movement
pressure and flow from waiting crowd
relative effectiveness of wide and narrow exits
other characteristics"
There are then numerous paragraphs explaining these factors, relative effectiveness of wide and narrow exits and other factors including a comparison of existing code figures for discharge rates. Hence the recommendation that the practical range of values for the discharge rate during escape would be about 30 to 45 persons per unit width per minute.
Remember the unit width was taken as 21 inches and the desired benchmark evacuation time was 2.5minutes.
21 inches equates to 533mm two units would therefore pass 80 person per minute for 2.5 minutes = 200 persons. Lo and behold 1066mm divided by 200 persons equates to 5.33mm per person.
Remember also that the actual flow rates measured in the various tests ranged from 19 to 107 persons per unit width per minute and that this cannot therefore be an exact science! But the system has proved itself over the years.
I believe where BS9999 is coming from is that if, because of early detection or a high ceiling etc the premises are actually safe for three minutes then the same 1066mm exit could pass 3 x 80 = 240 persons in that time or alternatively for 200 people you would only need to provide 4.4mm/ person.
One major difference between BS 9999 and previous studies is that all previous works were careful to specify that such interpolation should only take place for exits of at least two unit widths. BS 9999 ignores such advice and allows such interpolation to be applied to any exit irrespective of width. In my opinion this is the most dodgy aspect of BS 9999.
