How many fighters are there available for recruitment in a given village? How powerful is the highest-level wizard likely to be in a given region? What level of charitable healing is a normal church able to provide a given townsfolk? In short, what are the presumed class/level demographics of an AD&D world?
This question was not resolved in any official way in 1st Edition materials; however, I found that certain clues allowed me to develop a spreadsheet which could answer these questions in a generalized way. The DM's Guide section on Henchmen (p. 35) gave estimates for the proportion of level-advanced humans, and the ratios of the different classes (various NPC encounter tables gave other, similar ratios). The levels of military leaders could be associated with the number of men they led (p. 30). The World of Greyhawk Glossography likewise gave proportions for high-level character levels (p. 16). If these numbers couldn't be perfectly interpolated, they did give a general shape to the population curves involved.
Ultimately, I found that the demographics generally fit these descriptions if, for each class/level population, there were 2/3 that number of characters at one higher level. This approximately matched the clues given above, which were not compatible with the strict suggestions of some authors that only half of a given class/level were advanced by one greater level (my system works out to about 1/2 the population at +2 levels higher), nor to the generous 3rd Edition assumption of half a population being at twice a given level (3rd Ed. DM's Guide, p. 140). (Consider a town with 20 1st-level fighters: the "strict" system indicates a leader of only 6th-level; my "moderate" system provides for the highest fighter to be 10th-level; and the "generous" system allows for a character of 16th or higher level in the same town.) In practice, my "advancement percentage" is class-adjusted to represent the greater danger that certain professions face: I settled upon Clerics: 62%; Fighters: 60%; Magic-users: 65%; Thieves: 60%; and Other: 70%.
The results can be seen in the accompanying spreadsheet. The top part summarizes certain demographics for each of the urban centers given in the DMG (p. 173). The main body computes average class/level populations for a total human citizenry as entered on the line under the boldface "Population". This is done by taking 1/100th of the global population (per DMG p. 35), dividing into the class ratios on row 21, and applying the advancement percentages discussed previously. Sums per-level (for all classes) are in column B. Sums per-class are in row 54, and below that, ratios and totals of "name level" figures. Likewise, total high-level magic-users are shown above on lines 13-14: this shows that on a continent of 12 million people, there are about 30 Magi and 20 Archmagi in existence (which is vaguely reasonable).
Other results are shown on the later pages. Page 2 (columns I & J) shows that fighter/follower ratios approximately match the figures on DMG p. 30, allowing for some smoothing. Column K shows the high-level ratios which resemble those on WOG Glossography p. 16 (mine are somewhat steeper categories, but as close as they'd get without skewing other source ratios). Columns M-P show that all clerics on the hypothetical continent can muster about 400,000 cure disease spells per month, maximum; given that something like 600,000 people will contract disease or infection each month (DMG p. 13), we see that clerical healing cannot even theoretically wipe out disease in our standard AD&D world (and that's before accounting for neutral or evil clergy, those who are insular or seek hefty gifts, otherwise busied or spell-using clerics, etc.). Page 3 basically shows how many magic items are possessed by this population by standard methods, and shows that there's a disturbingly high number, given the rarity of wizards and the difficulty of magic construction in 1st Ed. (which is a point of advantage to the later 3rd Ed. rules).
One thing that should be pointed out is that my method diverges from the level-advanced proportion given on DMG p. 35, of 1:100 humans being able to advance in level. I have about twice as many (1 in 40) with greater than 0-level; obviously this is due to my taking 1:100 population for 1st level, and then adding in others at higher levels. This could be corrected by changing the initial division to 1:280 (in cell B22) and increasing advancement percentages by a few decimal places - and the column K numbers come out even closer to WOG Glossography p. 16; however, this creates too many high-level characters, as compared to the ruler levels on Glossography p. 17 (national rulers usually being in the 14-19 level range) and other sources like the characters in the high-level WG module series. Furthermore the Magi/Archmagi ratio starts to flatten out, with the latter threatening to outnumber the former, which would be counterintuitive.
Finally, this method can be tweaked slightly for other races, most of whom do not have any 0-level characters in their populations. In these cases, line 22 would have no "men-at-arms", different class percentages as appropriate for the race, and most of the population in the "level 1" category. Advancement would proceed in basically the normal fashion for higher levels.
Appendix
View a Microsoft Excel spreadsheet for human demographics here.View a rich-text version of this page here.