I've made some updates to my Subway + Frequent Bus map
. For review, bus routes and corridors in gold are routes that have rapid transit (ish) frequencies all day
, while the routes in bronze are high frequency during peak hours but not off-peak
. If you want to read more about that in absurd detail, you can check out my posts from September
First, I found a new bronze corridor
that I'd missed the first time around: Belgrade Ave & Center St in Roslindale and West Roxbury
. The 36 alone is 15 or 16 minutes at peak, and then the 35 and 37 both layer on 15-25 minute headways as well. Yes, it's possible that those headways are not evenly spaced (i.e. a 35, 36 and 37 arrive one after another, a minute apart), but unless they are literally right on top of each other, you're still probably looking at 12-min-or-better, so it seemed worth mapping. Worth highlighting that proper spacing would not only bring the peak headways to 7 minutes, but actually could even bring the off-peak headways to 14 minutes; though somewhat marginally so, that could possibly be enough to merit a Gold categorization. (This is also the unusual corridor where the PM peak is higher frequency than the AM peak.)
Second, I explicitly calculated the cumulative frequencies on all Gold trunks,
such as Roslindale Village-Forest Hills, Kenmore-LMA-Ruggles, and Wellington-Everett. I also listed out the individual routes which feed into each trunk.
Finally, I spent more time digging into what I call the "Dorchester Lattice
": the criss-cross network of routes that, unlike most of the rest of the system, don't fit neatly into a "radial vs circumferential
" framework. As an example -- the 66 is clearly a circumferential route, in that most journeys on it are probably not ultimately destined for Downtown: you're not gonna ride from Harvard to Roxbury Crossing and then hop on the northbound Orange Line to Downtown -- you'll take the Red Line instead. Likewise, the 73 is clearly a radial route -- you're not gonna use the 73 as a "shortcut" to travel more directly to your destination.
The Dorchester Lattice is filled with routes that defy this classification
. The 22 is a great example: on the one hand, it's circumferential in that it allows you to travel from Ashmont to Ruggles without going downtown; on the other hand, it also allows you to travel (say) from Grove Hall to Downtown, via a transfer to the Orange Line, and is one of several routes that feed into Jackson Square, Roxbury Crossing, and of course Ruggles.
Other routes are more clearly radial in nature, but once again there is a twist: there is an unusually high level of overlap between three distinct subnetworks
, centered on Nubian/Ruggles, Egleston/Jackson Square, and Forest Hills. (Who said the El was dead?) This isn't unprecedented -- Everett for example has overlapping Wellington and Sullivan networks. But it's more extreme here, all the more so because there are more high-frequency routes. You basically have three northern anchors
(the old El stations listed above) and two southern anchors
-- Ashmont and Mattapan -- and you pretty much have high frequency routes individually connecting each possible pair.
(The 21 -- Ashmont-Forest Hills -- is a bit marginal in its frequencies, but the principle still holds.)
So what have I added?
First of all, I added "cumulative corridors" to explicitly call out the high frequencies on those trunks feeding into those northern anchors
. These aren't easily clickable on the map, but can be accessed from the list of lines in the sidebars. These reveal the very
high cumulative frequencies on these stretches, unrivaled anywhere else in the network, by a wide margin. Listed below, with AM peak, midday, and PM peak headways in parentheses:
- Ruggles-Roxbury Crossing (2 min, 3 min, 2 min)
- Roxbury Crossing-Nubian (2 min, 3 min, 2 min)
- Nubian-Grove Hall (3 min, 5 min, 4 min)
- Jackson Square-Franklin Park Zoo (4 min, 9 min, 4 min north of Humboldt Ave, 5 min, 12 min, 5 min to the south)
- Franklin Park Zoo-Harambee Park (3 min, 6 min, 4 min)
The Nubian-Ruggles jog isn't totally unusual -- Forest Hills, Sullivan and Lechmere all have similar stretches where the last mile is just brimming with buses. But that stretch down Warren St to Grove Hall
? That's close to 2 miles of over 20 buses per hour
getting shoveled through during the morning peak. Outside of the Dorchester Lattice, there's nothing else like that in the system. Moreover, south of Grove Hall, the frequency on Blue Hill Ave drops because the 19 and 23 peel off... but they get replaced for the stretch down to Harambee Park with the 22 and the 29, which also doesn't happen elsewhere in the network.
This is detailed more in the respective Better Bus Profiles, but this also creates the scenario where a rider along Blue Hill Ave might make a "game time decision
" as to whether they will transfer to the Orange at Jackson or at Ruggles/Roxbury Crossing, depending which route's bus arrives first. Again, as far as I can tell, that doesn't happen anywhere else (except mayyyybe on the North Shore -- "Am I gonna ride the bus all the way to Haymarket today or will I transfer at Wonderland? We'll see which shows up first!").
There's another thing I learned, that doesn't directly show up on the map. See, I had been wondering, "did I miss any Bronze cumulative corridors in Dorchester the first time around?
" As it turns out, I did not -- and for a very interesting reason.
When plotted out against each other, Dorchester's bus routes fall into three very distinct buckets
- The 10-minute network: routes which have a peak frequency under 10 minutes.
- These routes all have midday frequencies of 15 minutes or better; most of them are 12-min-or-better, and the majority are under-10-min -- meaning, they are consistently high frequency all day
- There is one special case: the 21, which is 15 min AM peak, 15 min midday, and 12 min PM peak; but still, that frequency, though lower, is consistent throughout the day
- The 15-minute network: routes which have an AM peak frequency of 15 minutes
- These routes are notable for their significant frequency reduction off-peak: most drop to 30 min and one drops to 60 min
- These routes tend to parallel high-frequency corridors, and tend not to feed into trunk corridors as much
- This network consists of (unique stretches italicized):
- The 19 (Ruggles-Grove Hall-Fields Corner)
- The 42 (Nubian-Egleston-Forest Hills)
- The 44 (Nubian-Humboldt Ave-Egleston-Jackson)
- The 45 (Nubian-Blue Hill Ave [north]-Grove Hall -- paralleling Warren St)
- The 17 (Fields Corner-Andrew -- a bit of a special case, 20 min peak, 30 min off-peak, but otherwise pretty similar)
- The 30-minute-plus network: routes with an AM peak of 30 min or worse
- These routes are mostly at hourly or frequencies off-peak, a significant drop
- These routes are... interesting. There is a bit of a trend of them being oriented toward Andrew, and they also generally do not feed into trunk corridors, and tend to be "loners" along their corridors.
- This network consists of:
- The 14 (Roslindale Village-Mount Hope-Franklin Park Zoo-Nubian)
- The 16 (Forest Hills-Franklin Park Zoo-Andrew and sometimes to JFK/UMass)
- The 18 (Ashmont-Andrew via Dorchester Ave)
- The 30 (Roslindale Village-Mount Hope-Mattapan)
- The 41 (Jamaica Plain [Monument]-Jackson Square-Nubian-Uphams Corner-JFK/UMass -- the stretch between Nubian and Uphams Corner is not unique, but all the other segments pretty much are)
Even though there are some edge cases like the 21 and the 17, for the most part these categorizations are very robust
, and the differences highly distinct.
If this is hard to visualize, I've made a map. Thick lines are the 10-minute network, color coded high freq to low freq red to green (my apologies if you're colorblind). Thin green lines are the 15, and thin blue and black lines are the 30+.
All of this, incidentally, means that there indeed aren't
any hidden "Bronze" corridors à la Belgrade Avenue. Routes are either Gold, or they are mid/low frequency loners
. For the most part, the non-Gold routes are pretty similar to suburban routes like the 51 (Reservoir-Roslindale Village-Forest Hills) or 59 (Watertown Square-Newtonville-Newton Highlands-Needham) -- a far cry from "a bus every three minutes on Blue Hill Ave".
Now, does any of all of this particularly change any of my analysis that I made back in September? For the most part, I don't think so.
Probably the biggest impact on my analysis is the impact of having the cumulative frequencies in Dorchester for comparison:
I'd originally made a point of calling out five various Gold routes for their unusually high frequencies; while it's true that they still are the highest frequency single
routes, the cumulative frequencies along Warren Street, Blue Hill Ave, and Columbus Ave blow them out of the water. Particularly given the length of the Warren Street corridor, and then additionally the various segments along Blue Hill Ave, it's worth highlighting it -- this isn't just the 1 mile segment of Washington Street south of Forest Hills, but well over four miles of high-frequency service, arranged in a topology that doesn't exist anywhere else in the system
As always, feedback welcome!