Last week I singled out Richard Layman for repeating the chestnut "Spain doesn't have the population density to support economically many of the lines, based on ridership." Again, let me make clear that Richard is far from alone in assuming that density is required to support transit, and that his blog is informative, insightful and well worth reading for urban issues. He was also a good sport in leaving a comment on my post; unfortunately all the comments made it clear to me how deep the idea is ingrained in our understandings of transit.
A number of people have addressed this issue before. Richard mentions Steve Belmont and linked us to a scan from his book (see also David Alpert's take). Alon mentions Gary Barnes and his concept of "perceived density" (PDF); the Austin Contrarian has his own idea of perceived density.
I had a great idea for a phrase, "density is not destiny," but like most ideas it turns out that someone's thought of it before you; in this case it was Paul Mees, and Jarrett has an interesting discussion. But all these discussions are frustratingly myopic, assuming that the competing road network is a constant force of nature beyond political influence.
Instead of looking at the concept of "density," let's look at "support." What does it mean to support a transit line? Is it complete financial self-sufficiency, as Germà Bel demands for the Spanish high-speed network? If that's the case, then very few transportation projects anywhere would qualify. Is it the simple existence of the transit line? Then Newburgh's three-line transit system would qualify, since it exists, but that's not a very enlightening criterion. Is it a certain threshold of mode share, as the discussion at Greater Greater Washington would suggest? That's more promising, but it's not all.
Let's bring in some Strong Towns thinking. Chuck Marohn looks at any transportation project and asks, what is the return on investment? And it turns out the answer is connected to density. The ROI for a street, bus line, train line or ferry dock, it turns out, is dependent on the benefits derived from that investment. If it's a government investment, it has a "public ROI" indicating the benefits accrued to the public, whether in the form of tax revenue or any other goal.
ROI is the benefits divided by the costs. In transportation, sewers, utilities and other public projects, the costs are spread out geographically, so the ROI depends on the density of the benefits. That is where density comes in.