Appendix – Mapping the Bus Network Redesign: Somerville’s Buses, Present and Proposed

As an appendix to my main post on the topic, I will detail a series of modifications to the Bus Network Redesign that I believe will address most of the concerns raised by the community so far, while still maintaining most of the design philosophy embraced by the Redesign team. 

While I am not sure that it has been stated explicitly, I believe the Redesign was undertaken with a “net zero” assumption – meaning the Redesigners assumed they could only work with the existing buses the system has today, and not increase the number of buses overall. As such, I am going to frame my suggested modifications through a reallocation lens: if I propose lengthening one route, I will attempt to balance it out by shortening another.

(I am assuming a level of fungibility here that may be optimistic; there are operational concerns that may mean that it’s not as simple as shortening one route and using the now excess buses to lengthen another. However, I believe that my suggestions are modest enough that those operational concerns could probably be mitigated with more detailed planning.)



In order to keep the net total requirements for buses on the routes in question unchanged, I’ll be measuring things in terms of “thirty-minute route-miles” – the number of buses required to run a stretch of service at the 30-min-or-better frequency tier used by the Redesign.

The high-freq 15-min-or-better routes will therefore be counted as “doubled” route-miles, and the low-freq 60-min-or-better routes will be counted as “half” route-miles. 

This isn’t a perfect system, but it should be reasonable enough for our purposes here. 

Frequency adjustment and infrastructure as fudge factors

One nuance that I am largely ignoring is that it is actually possible for us to proverbially add two plus two to get five. 

Imagine we have five high-freq routes, that each run 6 bph to yield 10-min headways. If we reduce each route to 5 bph and 12-min headways – still well within the 15-min high-freq range –, then we have freed up an additional 5 buses, which we could then run on a whole new route, also at 12-min headways. This could be done without reallocating any route miles, as I am doing here.

Since we haven’t been given specific frequencies for all the proposed routes – just a bucket of “15-min-or-better” – I haven’t been relying on frequency adjustments to create a surplus, instead using only route mile adjustments. In reality, we might have a bit of a fudge factor available to us for the actual implementation. 

In the background of all of this also is a looming “infrastructure fudge factor”: many of these routes – especially the long ones – will require dedicated bus lanes and transit priority signaling to be effective. So that also makes our calculations somewhat imprecise, but still the best we can do for the moment.

Generating a surplus

Shortening the proposed 87: +3.33 route-miles

I suggest shortening the proposed 87 significantly, terminating it at/near the intersection of Mystic Ave and Harvard St. EDIT: The segment from Arlington Center to Turkey Hill would be transferred to the Redesign’s 90.

Savings: 3.33 route-miles (not including the Turkey Hill segement, which is immediately transferred to the 90 and not available for the surplus)

Loss of service relative to current system: None; the unique stretch of the proposed 87 runs along Harvard St and has no current service

Loss of service relative to the Redesign: Crosstown feeder service to Davis is lost, though alternative service on the T96 is available less than a half-mile away 

Justification: The proposed 87 aims to achieve the following four purposes simultaneously, in order of efficacy:

  1. Serve Mystic Ave as feeder service to Sullivan
  2. Provide new crosstown feeder service to Davis
  3. Serve Clarendon Hill and northern Broadway as feeder service to Davis
  4. Serve Clarendon Hill and Davis as crosstown service to Sullivan

Objectives 3 and 4 can be achieved by other means, and Objective 4 in particular can be achieved more effectively with a less roundabout route. Objective 2 is admirable, but is net-new to the system and thus lower priority if ranked against existing established travel patterns. That leaves Objective 1, which can be achieved simply and reliably with a feeder route that terminates at Harvard St.

Rerouting and shortening the proposed 90: +2.85 route-miles

I suggest rerouting the proposed 90 to run along Somerville Ave and terminate at Union GLX instead of Wellington via Highland Ave and Assembly. 

Savings: 2.85 route-miles

Loss of service relative to current system: Loss of one-seat bus to Assembly and crosstown service to Sullivan from Highland Ave (currently running every 35-40 minutes for most of the day); both can be mitigated, as discussed below

Loss of service relative to the Redesign: Loss of one-seat bus to Assembly from Highland and from Wellington, and loss of crosstown service from Wellington; both can be mitigated, as discussed below

Justification: As discussed in the main post, the Redesign takes the existing frequencies of the Highland and Somerville Ave corridors and swaps them. Redirecting the 30-min corridor down Somerville instead of Highland aligns better with current frequencies.

Critically, rerouting one of the Clarendon Hill routes down Somerville Ave maintains the existing Clarendon – Davis – Union service pattern that has been in place for over a century. Preserving this service pattern makes this reroute a high priority.

Final total surplus: 6.18 route-miles

EDIT: An oversight in my original analysis did not consider the Redesign’s proposed stretch of the 90 to the east of Wellington, continuing on to Chelsea. The Wellington-Chelsea segment remains untouched in my suggested revisions, though I would also suggest some tweaking of the northeastern network in order to reroute the Chelsea-Wellington route to become Chelsea-Sullivan instead, taking advantage of the transfers at Sullivan, as well as my proposed high-freq shuttle service to Assembly.

Utilizing the surplus

Rerouting the T39 to Highland Ave and extending it to Davis and Clarendon Hill: -3.6 route-miles

I suggest rerouting the proposed T39 along Highland Ave instead of Somerville Ave, extending to Davis and beyond to Clarendon Hill. This would essentially supercharge the current 88 route, rerouted to Union, and through-run to Cambridge and Longwood.

Requires: 3.6 route-miles (1.8 miles of distance, doubled for high-freq)

Loss of service relative to current system: Essentially none; current 88 journeys that end at Lechmere without transferring to the Green Line would now require a transfer at Union Sq GLX

Loss of service relative to the Redesign: Essentially none; crosstown service to Sullivan likely can be maintained at mid or low frequencies, as available in current system, via a separate route described below

Additional service relative to current system: Enormous increase in one-seat access from Highland Ave, Davis, and Clarendon Hill, including to Cambridge and to Longwood; plus all benefits listed below relative to the Redesign

Additional service relative to the Redesign: Creation of one-seat service from Highland Ave to Union Sq, and overall increase in frequency of service to Union Sq from Clarendon Hill and Davis when combined with Somerville Ave service; maintenance of strong Green Line connectivity for Clarendon Hill, Davis, and Highland Ave (essentially by relocating Lechmere transfers to Union GLX)

Justification: As discussed above, running the high-freq service along Highland instead of Somerville Ave aligns with current frequency tiers, and thus preserves long-running existing travel patterns. Extending service south of Union then creates a strong axis of transit service running across Somerville and eastern Cambridge that provides feeder access to both the Green Line and Red Line, while still offering unique service.

Moreover, running Highland service out of Union GLX provides unprecedented additional access to Union Square, which helps mitigate Somerville’s lack of north-south bus service (difficult because of hills); running out of Union GLX also reduces the need for an otherwise-awkward diversion to serve Gilman Sq GLX door-to-door, offering alternative door-to-door transfers at Union instead. This reroute better integrates to and complements GLX than the current Redesign proposal does. 

I would argue that this reroute is the most urgent change required in the Redesign proposal.

Remaining surplus: 2.58 route-miles


Rerouting and extending the T39 to Davis offers a few opportunities for variation. 

I believe an extension to Clarendon Hill is necessary, but truncating at Davis reduces the demand by 1.8 route-miles (which we get by taking the 0.9 miles of distance and doubling it in order to determine the equivalent “30-min-or-better route-miles”).

A more direct route between Davis and Union would travel via Summer St, and then cut over to Highland Ave at School St/Prescott St, reducing the demand by .4 route-miles. However, I am unsure whether the grades on that alignment are too steep, so an alternative alignment would run via McGrath Hwy and Washington St. I have used the more conservative McGrath alignment for the figures here.

Institute a Clarendon Hills-Davis-Highland-Sullivan route: -2.0 route-miles 

There is one gap remaining, originally covered by the Reimagined 90: Clarendon/Davis to Highland Ave, jumping over to provide crosstown service to one or more Orange Line stations. 

We can break this up into 3 segments, totaling 6.4 miles:

  • 4 miles from Clarendon Hill to Sullivan (3 miles if only Davis to Sullivan)
  • 1 mile from Sullivan to Assembly
  • 1.4 miles from Assembly to Wellington

In terms of system coverage, the following journeys remain poorly supported by my suggested revisions to the Redesign:

  1. Clarendon Hill to Sullivan Sq
  2. Davis to Sullivan Sq
  3. One-seat bus access to Assembly from Somerville
  4. Two-seat bus access to Assembly from Wellington bus destinations

Journey category 4 is pretty poorly supported by the Redesign as it stands; the proposed 90 sits in the 30-min-or-better category, which will not be friendly to bus transfers. As such, I’ll consider it out-of-scope for the moment. 

Category 3, I would argue, should be deprioritized as a one-seat ride, in favor of a frequent and reliable shuttle service from Sullivan, as discussed below.

So that leaves the crosstown service from Clarendon/Davis to Sullivan – 4 miles. At 30-min frequencies, we can immediately support 2.58 of those four miles. If we reduce frequencies into the 60-min-or-better territory, then we can easily stretch the remaining surplus to support one-seat rides between Clarendon, Davis, Highland, and Sullivan.

As it happens, this is consistent with the service provided by the current 90 bus and in fact would still be a service enhancement, as there would be sufficient surplus to extend this route to Sullivan; this would restore the single-ride between Clarendon and Sullivan which the Redesigned 87 originally did (but which is lost with my suggested revision to the 87), and would do so more directly. 

Serving Assembly

By the Redesign’s own philosophy, transfers to a frequent service are sometimes more valuable than a one-seat ride. The current Redesign proposes one route to serve Assembly (by bypassing Sullivan). I suggest that centralizing routes at the hub at Sullivan and providing a frequent shuttle between Sullivan and Assembly is a more effective option.

Frequent service from Sullivan would mean that riders from all Sullivan bus routes would benefit, not just the odd one that could be extended at low frequency to Assembly. Shuttle service would also relieve pressure on the Orange Line for journeys between Sullivan and Assembly, and could be configured to offer better door-to-door service directly to stores – valuable both for passengers with mobility impairments as well as those with shopping items.

A circulator bus route from Sullivan to the Assembly development could run end to end without stops and without traffic in 12 minutes, according to Google:

Almost all of the streets on the above route are wide enough to support bus lanes, meaning buses could be largely unencumbered by traffic. A pair of buses would almost certainly be able to provide all-day service at least every 15 minutes and three buses would likely offer headways better than 10 minutes.

Centralizing transfers at Sullivan and providing connections to Assembly via high-frequency high-reliability shuttles would provide many more riders with bus access to Assembly Row. 

I would also suggest that Sullivan-Assembly shuttles potentially could be partially subsidized by companies located at Assembly Square. Mass General Brigham – which I believe remains the state’s largest private employer – has headquarters at Assembly Square, for example. It would be vital that these shuttles remain integrated into the T, both from a scheduling and a fare perspective, but private employers could, for example, contribute to cover the costs of 3 or 4 new mini-buses to expand the T’s bus fleet to enable this service. Free transfers to the shuttles would be critical.


Debating one-seat vs two-seat bus access brings up an issue I haven’t discussed yet: fares. Many of the Redesign’s proposals require adding rapid transit fares into journeys which currently are bus only, resulting in higher costs for passengers. If the distinction between rapid transit fares and bus fares were meaningless, transferring to the Orange Line becomes a more feasible option for low-income riders. 

This topic deserves a post of its own, but the short answer is that the T should acknowledge that, if you are affluent and can afford a monthly pass, the T is cheaper than if you are poor and have to pay each fare individually, and attempt to adjust its fare system accordingly.

Additional “levers” for adjusting the system

These are not changes that are required to enable my suggestions, but they may be useful as additional “levers” for adjusting the system overall.

Shortening the proposed T96: +1.3 route-miles

I suggest shortening the proposed T96, terminating it at Davis instead of Porter.

Savings: 1.3 route-miles (the actual distance is 0.66 miles, but run at 15-min freqs, so therefore doubled relative to 30-min-freq routes)

Loss of service relative to current system: Loss of half-hourly one-seat from Medford Sq and College Ave to Porter and Harvard stations

Loss of service relative to the Redesign: Loss of high-freq one seat from Malden, Medford Sq, and College Ave to Porter station

Justification: The proposed T96 is a classic trade-off for the Redesign: the route is extended to the north, frequency is increased significantly, and some additional transfers are required. The majority of the current 96’s riders alight at Davis; comparatively few alight at Porter, and a large fraction alight at Harvard. Losing the one-seat to Harvard is controversial to be sure, but if you are going to do it, the data suggests it would not be a huge loss to also lose the one-seat to Porter, which appears to be a noticeably lower priority. As a high-freq route, shortening the T96 gives you “double bang for your buck.”

Reroute the T101 to directly serve Ball Sq GLX: -0.66 route-miles 

I suggest diverting the T101 to continue along Broadway to provide a door-to-door transfer with Ball Sq GLX, turning north on Boston Ave, east on Harvard St, and returning to the original alignment on Main St.

Requires: 0.66 route-miles (0.33 miles of distanced, doubled)

Loss of service relative to current system: Loss of “hypotenuse” service on Main Street south of Tufts Park (already proposed in Redesign)

Loss of service relative to the Redesign: The “hypotenuse” service was originally replaced by service on nearby Medford St; this diversion puts that replacement service farther away

Additional service relative to current system: Provides feeder service to Ball Sq GLX on Harvard St and northern Main St; increases frequencies on Broadway all the way to Boston Ave

Additional service relative to the Redesign: Same as with the current system, and provides stronger connectivity to Ball Sq GLX than current proposed; this reroute would also partially mitigate the shortening of the proposed 87 by offering an “east-west” feeder service running into a rapid transit station – Ball Sq GLX replacing Davis Sq

Justification: This one, I am somewhat undecided about. 

On the one hand, general best practice would call for a major thoroughfare like Broadway to be anchored at both ends by door-to-door transfers to rapid transit. Even though journeys to Ball Sq GLX would require some back-tracking, the overall journey time and convenience likely would be improved, relative to spending a longer time on the bus. 

Moreover, for Broadway passengers that need to get to Green Line (e.g. because they work near Hynes Station), boarding the Green Line directly will likely be more convenient than riding all the way to Sullivan, changing for Orange, and changing again at North Station. Providing access to the Green Line would also somewhat help mitigate the lack of “north-south” bus routes in Somerville, insofar as GLX is slightly more north-south than the bus routes.

On the other hand, this diversion does leave the “hypotenuse” section of Main St with significantly less access. Moreover, the northern half of the T101 beyond Ball Sq duplicates the T96’s feeder service into Medford/Tufts, so southbound T101 riders might see less benefit. (Although, this diversion would also double the number of buses running from Medford Square to the Green Line, meaning “opportunistic” riders looking to transfer to the Green Line anyway will have twice as many choices – so there would still be benefit.)

And the question would be, how many Broadway riders are looking to do that reverse-jog to access Ball Sq GLX?

One way to reduce the impact of this question would be to extend my suggested shortened 87 down Main St and Medford St, to terminate at Magoun Sq GLX, Gilman Sq GLX, or Union Sq GLX. That would at least provide the “hypotenuse” section of Main St with compensating feeder service to rapid transit. (See below.)

Ultimately, I do think that anchoring Broadway with rapid transit connections at both ends is a compelling enough reason to put forward the suggestion, even if I still view it as more provisional.

Extending the Somerville Ave bus to Sullivan: -1.0 route-miles

My suggested reroute of the proposed 90 runs Clarendon Hill – Somerville Ave – Union GLX, every 30 min or better. Instead of terminating at Union GLX, this reroute could instead carry on along Washington St, connecting with the Green Line instead at East Somerville GLX, before terminating at Sullivan.

This would maintain Clarendon’s half-hourly connection to the Orange Line, and would create new connections between Porter and Sullivan; it would require using 1.0 route-miles out of the surplus.

Extending the Mystic Ave bus south to GLX: between -1.0 and -2.5 route-miles

If there is additional surplus available, the proposed 87 – which I have suggested curtailing at Harvard St – could then be extended back down Main St to terminate at one of the GLX stations (e.g. Magoun Sq GLX, Gilman Sq GLX, or Union Sq GLX). This could provide improved north-south connectivity across the city, and would do so along streets currently served by similar frequencies.

A brief note on the T109 and 86

Currently, the 86 runs from Sullivan to Harvard to Reservoir. A significant fraction of 86 riders travel through Harvard from one side to another. According to my analysis, a typical westbound 86 bus enters Harvard station with about 1,100 passengers; about 500 disembark – meaning about 600 are continuing on to Allston/Brighton. That’s a significant chunk of ridership – over a quarter of passengers bound for Allston/Brighton board before Harvard. (Another quarter board at Harvard.)

The Redesign makes a bold proposal: extending the 109 (increased to high-freq service all day) from Sullivan to Harvard, enabling a one-seat ride from Everett.

The downside is that the 86 is truncated at Harvard, requiring riders to transfer to reach Sullivan. Given that Sullivan is a transfer hub, it’s likely that at least some of those riders have a second bus they need to transfer to – which could turn such journeys under the Redesign into three-seaters.

In general, the Redesign has avoided short-turn services and through-running (where a bus arrives at its “terminus”, changes its rollsign, and continues on a second route). However, I believe this would be an area where some combination of overlapping short-turn services and/or through-running would be worthwhile. Perhaps every other T101 through-runs Harvard and continues as an 86, essentially making the T101 into a nominal short-turn service of an integrated T101+86. While this route would be quite long, it may be feasible if sufficient reliability infrastructure is in place. 

(Alternatively, depending on how the Somerville routes get reshuffled and what frequency fudges are available, it might be possible to free up enough buses simply to extend the 86 on a thirty-min-or-better basis to Sullivan full-time. That could be a valuable relief valve if the T109’s start bunching up inbound in Everett – short-turn one or two T109s at Sullivan, and have the 86s available to pick up the slack and keep frequencies high between Harvard and Sullivan.)

A brief note on the T39

EDIT: This topic possibly merits a post of its own. I have left the T39 itself untouched aside from the reroute and extension described here. However, I believe the T39 requires further review overall. The very high frequencies required in Jamaica Plain likely would probably be a bit more than required for either potential corridor in Somerville. I also believe that the existing Porter-LMA-Forest Hills proposal is possibly too long to be reliable and (most importantly) severs a heavily used link between Jamaica Plain and the eastern end of Huntington Ave. As with the T101, the T39 may benefit from short-turns, or overlapping routes that create high frequency cores.