100 doesn't represent a physical number. It's just an indication of the relative frequencies. If, for example, Manhattan had typically two fixies for sale per person (this is completely made up), then Brooklyn would have 1.25.
Yes, per capita literally does mean "per person", but the purpose it serves here is to allow meaningful comparison of more than just raw numbers. Manhattan might have 1,000,000 fixies, and Brooklyn might only have 100,000 fixies. This by itself doesn't really tell you much, since what if Manhattan had 1,000,000 people and Brooklyn only had 50,000 (again completely theoretical); then, the "per capita" fixie bike index would be twice as high in Brooklyn.
"Fixies per person" in a geographic region isn't a "raw number".
I don't get your point with the example, for those values I get "fixies per capita" of 1 in Manhattan, and 2 in Brooklyn. That clearly states that there are twice as many fixies per person in Brooklyn as there are in Manhattan.
If you also normalize based on Manhattan, you'd get 100 and 200 which say exactly the same thing but with more zeroes and no longer being "per capita".
Sorry if that was unclear. The original comment stated that using per capita in this manner would be "inaccurate", but you can't compare Manhattan and Brooklyn simply by comparing the total number. What I was trying to explain in response was that Priceonomics didn't have "fixies per person" data, only the total fixies for sale in a region (the "raw number"). Without using per capita, these results could easily be misconstrued.
Yes, per capita literally does mean "per person", but the purpose it serves here is to allow meaningful comparison of more than just raw numbers. Manhattan might have 1,000,000 fixies, and Brooklyn might only have 100,000 fixies. This by itself doesn't really tell you much, since what if Manhattan had 1,000,000 people and Brooklyn only had 50,000 (again completely theoretical); then, the "per capita" fixie bike index would be twice as high in Brooklyn.