Portland: Counting by 17

Frequentservice portland As hard budget shortfalls sweep across North America, transit agencies are making all kinds of changes to balance the budget.  Portland’s Tri-Met tried at first to cut low-ridership services, but as the red ink keeps flowing they’ve finally had to cut something every urbanist should care about.  They’re cutting the core Frequent Network, the service that’s designed to meet the needs of people who want to get around the city easily all day, with spontaneity and a sense of personal freedom.

Tri-Met was one of the first North American agencies to define and promote a Frequent Network, a clear way for customers to identify the major lines where a bus or train is always coming soon.  At the time, they defined the network as having frequencies no worse than every 15 minutes all day, every day.  They published a simple diagram of the network (above, click to enlarge) and also highlighted it on their network maps.  One theme of my professional work has been to encourage other agencies to follow their example.  Gradually, the idea has spread, from Minneapolis to Los Angeles to Adelaide; private mapmakers have also picked up the theme.  Los Angeles upped the ante: their Frequent Network is called the “12 minute network”, because it’s that frequent or better all day.

Now Portland’s 15-minute standard is slipping to 17, and sometimes 18, and sometimes even 20.

In Norton Juster’s great children’s novel The Phantom Tollbooth, a bombastic fool called the Humbug had the habit of answering every question by shouting “Seventeen!”  It’s a funny number.   Like most prime numbers, 17 carries a slight whiff of absurdity and subversion, expressed most directly in the impossibility of doing any arithmetic with it in your head.

Remembering the schedule is just the most obvious problem.  Can you remember the schedule pattern of a 15-minute service?  Easy.  All I remember is the first departure time of the hour, say :05, and I can derive the whole pattern whenever I need it, and it’s the same every hour.  If I know the service goes at 1:05 I also know it goes at 2:50 and 3:35.  Now try that with a service running every 17 minutes.

Tri-Met is not the first agency to flirt with 17, nor is this the first time.  In fact, I remember working in their planning department doing service cuts around 1984.  We talked about this very possibility, shuddered at the prospect of asking people to count by 17, and set it aside.  And we certainly hoped that in a city as pro-transit as Portland, the next generation would be fighting different battles.  We wanted to say:  “Son, we counted by 17, so you’d never have to.”

Tri-Met is doing what they can with what they have, and their planners did what they could to avoid this moment, but it’s here.  Here’s the stark fact.  In the grid network covering the densest parts of Portland, major transit spines like Belmont and Hawthorne will soon be less frequent than they were 25 years ago, even though the inner-city population they serve has grown massively in that time.  How could America’s self-proclaimed transit paradise have let that happen?

30 Responses to Portland: Counting by 17

  1. Jennifer November 4, 2009 at 7:27 am #

    Fiscal instability comes for everyone, even Portland. You pretty much answered your own question, though: Exploding population puts a strain on a transit agency’s operating funds and, like most other struggling agencies, operating funds are harder to come by than capital funds. If operating funding increases don’t keep pace with expansion and usage, eventually service will have to be cut.
    As much as Portland prides itself on its public transit, I can’t see that the public will allow a great service like this to deteriorate. I’m sure they’ll think of something.

  2. Adron November 4, 2009 at 8:51 am #

    Excellent write up Jarrett.

  3. t.a. barnhart November 4, 2009 at 8:57 am #

    given that most buses run late anyway, the 15-minute service is more cosmetic than anyway, at least in terms of accurate planning. i still head to the stop 10 minutes early & plan to wait. nearly 30 years of TriMet experience has taught me to expect anything but adherence to a schedule.

  4. EngineerScotty November 4, 2009 at 10:22 am #

    A few things particular to Tri-Met:
    * The agency made a bad bet in the oil futures market during the recent shortage, trying to insulate itself from higher gas prices, and as a result is buying gas for its busses at a higher price than you or I pay. This should expire eventually, or at least even out when the economy recovers and gas goes up for everyone else.
    * WES, the commuter rail service in Washington County, has not performed well at all. The service–a short (15 mile) commuter rail line using FRA-compliant DMUs on an active branch line, costs mucho $$$ to operate, and attracts few riders. Tri-Met is probably obligated to operate it at least for a while longer, as the project was federally funded.
    * The expansions of MAX have generally been in addition to bus service, rather than replacing it (other than entirely redundant lines); however, when you do this you stretch the operating budget further. When tax revenues go down…
    * Reportedly, Tri-Met will bond some of its payroll tax receipts to help build Milwaukie MAX–which strikes me as a particularly bad idea. Supposedly, Tri-Met actually has bonding authority (granted via an earlier bond measure) to issue tax-backed bonds to pay for this, but for some reason doesn’t want to use the bonding authority for the Milwaukie line.
    Needless to say, all of this is damaging Tri-Met’s reputation, especially among bus-riders, who are more and more calling into question Tri-Mets increasing forays into rail. (Insults like “railfan” get thrown around quite a bit). An unfortunate “us vs them” dynamic seems to be developing, similar to what has long been seen in LA–despite the fact that MAX is not a “nice neighborhood only” service.
    Regarding the 15 vs 17–the hope is that this is a temporary service cut until the economy picks back up.

  5. al m November 4, 2009 at 2:45 pm #

    That was a very interesting perspective!
    Some of the comments above nicely express some of the points I was going to make so I won’t bother repeating them.

  6. Daniel November 4, 2009 at 6:17 pm #

    Hmm, what’s the likely effect on revenue from service cuts such as these?
    If you’re cutting service from every 15 minutes, surely you’d consider just going to 20, rather than something so unmemorable like 17? (Or perhaps going to 20 on some routes so you can stay at 15 on others?)

  7. Pantheon November 4, 2009 at 7:07 pm #

    The futures market speculation on crude oil was not only disastrous, it was extremely irresponsible. Throughout the multi-year climb in the price of crude, only one airline (Southwest) employed a hedging strategy. For years, Southwest management was praised as brilliant, and analysts wondered why the majors didn’t follow suit. We all found out exactly why when the price of oil crashed. Southwest started reporting losses while the majors moved to profits. The problem with hedging is that it really is just speculation, except in limited circumstances that aren’t relevant for me to detail. I suppose Portland sees itself as sort of like the municipal equivalent of Southwest Airlines – the unassuming little brother to the majors, who goshdarnit does things differently and better. So Trimet saw Southwest reporting profits even as the price of oil climbed, and figured they ought to follow a similar strategy. But there is a difference: it is one thing for a private corporation to use shareholders’ capital to speculate on commodity prices, but quite another for a public agency to do the same with taxpayer dollars. This was a dereliction of duty on their part to both taxpayers and transit users, and every member of the Trimet board should be held responsible and terminated for their folly.
    Given this financial predicament however, they probably have little choice but to cut the frequent service lines. As a matter of simple mathematics, there’s only so much money you can save by cutting obscure bus routes that only run six buses a day. Sooner or later they had to cut something real, and they did. That said, there is something rather discordant about the way that Trimet does business. They are cutting bus services even as they expand the MAX system and plan a major expansion of the streetcar system. There are two problems here: first, the MAX serves surbanites and the bus system serves the people of the city, and second, the streetcar is perceived as a development or marketing tool (a gimmick, if you will) while the bus system is the lifeblood of the system for most transit users. What the poster above said about an “Us vs. Them” mentality developing is absolutely correct, and is likely to continue to worsen.
    I wouldn’t necessarily mind an expansion of the MAX and streetcar at the expense of the bus system if the new investments were logical from the position of making mobility improvements. However, in many instances these investments result in NO MOBILITY IMPROVEMENT.
    The streetcar results in no mobility improvement for reasons stated in your prior streetcar post, there is no reason for me to repeat it as you said it better yourself. And note that the proposed streetcar expansion runs along the same routes that are already served by buses (i.e. Belmont). So they are spending money to do the exact same route they are already doing, with no mobility improvement, but with a different technology. And this at a time when money is so scarce that they are cutting service along the main bus arteries.
    The case of MAX is more complex. I myself am highly critical of the system (with the exception of the Blue line running west of downtown) because I apply a simple test that I have learned to apply from reading this website: does it improve mobility? The whole point of having high speed trains is that they are supposed to go faster than buses. If they don’t, you have to question why you have trains. It has been said that Portland doesn’t have the money to build the kind of super-premium line that Seattle has with its “central link”. But to my mind, the only reason to build a train system is to make a mobility improvement over what buses can do. If you have the money to build a great train system, then build it. If you don’t have the money, then don’t build it. Add to your bus service instead.
    Here is an example of what I am talking about. A while back we had a discussion over the efficacy of putting light rail next to freeways. I was generally against it. Now I will admit, living next to a freeway might be alright if you also get walkability to a light rail station as part of the package, IF AND ONLY IF the access to the light rail station provides superior MOBILITY. But in the case of the I-205 (green) line, there is no such benefit. And if there is no mobility benefit, then the tradeoff of living next to the freeway doesn’t make sense. And if it doesn’t make sense, then people won’t want to live there. And if people don’t want to live there, then there won’t be any development. And if there is no development, then the whole thing is a failure – a train serving lonely, empty stations along a freeway. They built it, but no one came.
    But what do I mean about no mobility improvement? Trains are fast, right? Only if you do it right. To save money, the green line goes all the way up to Gateway and around, wasting precious time.
    Here are some case studies pulled right from TriMet’s timetable:
    You live next to the MAX station at SE Foster & 92nd. This also means you live right next to the #14 bus line. The #14 will take you to the heart of downtown in 39 minutes. The MAX gets you there in 36 minutes. That is a 3 minute advantage for the MAX.
    You live next to the MAX station at I-205 and Powell. #9 bus: 35 minutes. MAX: 32 minutes. Advantage: MAX by 3 minutes.
    You live next to the MAX station at I-205 and Division. #4 bus: 30 minutes. MAX: 31 minutes. Advantage: bus by 1 minute.
    Note that these comparisons assume the best possible scenario for the MAX. I am assuming that you live right next to the MAX station, which is also right next to the freeway. But the advantage of the bus is that they go into real neighborhoods where people actually live. What a concept! I know how revolutionary that must sound, to put transit services where people actually live. But bear with me here. Anything other than living right next to the freeway means the bus is significantly faster than the green line MAX.
    Let’s say I live at SE Holgate and 60th. Not exactly an expensive place to put up a condo building. Here are my choices: #17 bus or combination bus and MAX. I can take the #17 bus directly downtown. Time: 32 minutes. Or I can take the #17 in the OPPOSITE DIRECTION to connect to the MAX. Time: 5 minutes. Average wait for the MAX based on rush hour frequencies of 15 min: 7.5 minutes. Time on MAX train: 34 minutes. Total: 56.5 minutes. Advantage: bus by a whopping 24.5 minutes!
    So to summarize: the MAX is equal to but no better than the bus in mobility if you live right at the MAX station which is right next to the freeway. Live anywhere else and the bus is much faster than the train. Why would you build train routes that serve nobody and are slower than buses, and cut the bus routes to fund them?
    I have no problem with cutting bus services if Trimet truly has no options. But they do in fact have options. It was their choice to speculate on commodity prices with taxpayer dollars. It was their choice to expand a streetcar system that goes the same places as buses at the same speed. It was their choice to build a train system that provides no mobility advantage over bus services, and as I have demonstrated is actually at a mobility disadvantage, and serves no one and is not likely to serve anyone in the future because the development will not come. Why would it given the mobility facts I have demonstrated?
    So to answer your question, which was how could a transit-oriented city like Portland find itself in this predicament, the reason is that TriMet is run by stupid people. Intellectually challenged, if you prefer. That’s really all there is to it.

  8. Pantheon November 4, 2009 at 7:32 pm #

    Sorry, I just realized I made a simple mathematical error above. In the SE Holgate and 60th example, it takes 46.5 minutes on the MAX, for a bus advantage of 14.5 minutes.

  9. calwatch November 4, 2009 at 8:18 pm #

    Actually, there is a good reason to do fuel hedging, and that is budget consistency. The converse is an agency like Salt Lake City UTA, which imposes “fuel surcharges” on its riders every time gas prices, go up above a certain level. Utah is lucky in a sense because their board is conservative and as such gave staff the authority to index fares based on fuel prices. Trimet, in blue Oregon, likely would have never received that authority.
    Sure, you save money in the short term if the economy crashes, but you have to remember that in 2007 and 2008, when oil prices zoomed above $100, it was an open question whether it would hit $150 or $200. Today, with oil prices hovering around $80, despite the global recession, I don’t see oil getting significantly cheaper. Certainly the era of $50 a barrel oil is oil, unless they find another Saudi Arabia in a stable country or the US dollar significantly strengthens.

  10. Alon Levy November 4, 2009 at 10:05 pm #

    Why should rising population cause financial stress to a transit operator, instead of windfall profits? In Calgary, the city’s population boom has led to a transit ridership boom, leading to light rail revenues 4-6 times higher than operating costs, which are now financing further expansion.

  11. EngineerScotty November 4, 2009 at 10:31 pm #

    As an aside–I prefer threaded comments (or the option, at least) to linear ones, if that’s possible.
    In response to Pantheon’s comments concerning the Green Line–the I-205 segment of the line appears to be geared towards park-and-ride commuters; there are lots of them in the Sunnyside/Happy Valley area who were driving downtown before, and now are parking and riding. Whether a line which serves suburban commuters in this fashion is a good idea or not, is an interesting topic of debate–on one hand it gets them off the freeways (and out of downtown parking lots); on the other hand, it doesn’t eliminate auto dependency–just a particular set of trips. The biggest weakness of the line (and of MAX in general) is that the long slow slog through downtown makes it less useful for crosstown trips. The Green Line is nice and fast until you get to Lloyd Center; then it becomes a bit painful.
    Much of the acrimony concerning Tri-Met’s recent management decisions, is the old suburban commuters vs inner-city residents question. All around the US, you can find examples of city residents, many of them transit-dependent, getting angry when the local transit agency builds suburban infrastructure–especially when it comes with inner-city service cuts. In some cities, this gets involved with politics of race or social class (less so in Portland). Frequently, rapid transit (usually rail) is what gets built to the suburbs (local bus service being inappropriate for this application), so bus vs rail becomes a proxy for the suburbs-vs-city debate.
    Of course, Tri-Mets service district (and tax base) encompasses far more than Portland, and numerous suburban communities, especially Washington County (which has a high-tech industrial base and thus contributes a significant part of Tri-Met’s payroll tax revenue), like to complain they don’t get adequate service from Tri-Met. Occasionally, they even threaten to withdraw from the service district, as the city of Wilsonville did many years ago. Most of the suburban communities lack the density to make quality transit service practical, but do want some service nonetheless. Much of the reason WES got built is because of Washington County’s complaints.

  12. EngineerScotty November 4, 2009 at 10:38 pm #

    To respond to Alon–the trains in Calgary are full most of the time, as driving there sucks. 🙂
    Driving in Portland only sucks FTMP during rush hour, or if you need to park downtown. While MAX trains are crushloaded during rush hour, the things run 17 hours/day, and many early morning and late evening trains are nearly empty. Yet the driver and the power company still get paid.
    Portland has been investigating things like congestion pricing or tolls in order to discourage driving; however right now these things are politically difficult (or downright illegal under current law) to implement. (You cannot erect tollbooths on an Interstate, for instance, unless part of a funding mechanism for new construction).
    Out of all Portland’s transit modes, MAX loses the least amount of money in the aggregate, and MAX trains can operate in the black when full. But they often are not full.

  13. calwatch November 4, 2009 at 11:30 pm #

    And part of it is due to political considerations. For example, you could shortline Red or Green Line trains at Gateway during weekends and off peak hours and save money. But that would downgrade light rail’s usefulness for those going to Clackamas or the airport. Instead, there’s an excessive 16 trains an hour between Gateway and Downtown Portland, seven days a week.

  14. anonymouse November 4, 2009 at 11:39 pm #

    My impression of Portland, from having been there, is that all the talk about being a self-proclaimed transit paradise really is mostly just talk. Even the fairly small downtown has lots of parking lots, and the rest of it is as auto-oriented as any other place in the US. Transit service is passable but not great. It doesn’t have nearly the mode share of NYC, and it doesn’t have the same level of mobility and accessibility as NYC, nor nearly as many people taking transit. And the same is true for bicycling too.

  15. Jarrett at HumanTransit.org November 5, 2009 at 12:00 am #


    These cuts are tempting because it's easy to imagine that, hey, who'll notice two minutes difference.  But the clock headway problem is a real issue for many regular riders.  I'd have preferred going to 20 on the weak lines to keep 15 on the strong ones.  But I'd really rather that Tri-met and the City of Portland were taking a hard look at their priorities.   

  16. Jarrett at HumanTransit.org November 5, 2009 at 12:07 am #


    To be fair, I meant that Portland is a self-proclaimed transit paradise among cities of its approximate size and age.  I just came from doing a presentation to city planning staff in Nambour, Queensland and even there everyone's heard of Portland.  But there's no point in comparing Portland to New York City. 

  17. Mizzz November 5, 2009 at 12:59 am #

    Thanks for the informative article. Too often, I feel, people laud Portland/Tri-Met for its forward thinking and don’t hold it accountable for the lack of usability. Such as cutting bus service that can cover larger areas, is cheaper to cut/add service and generally faster. The MAX trains are user friendly and the ride is much nicer but who cares if I still need a car. I work until 12:30am and am unable to take the MAX because the last stop is 11:40pm.

  18. EngineerScotty November 5, 2009 at 1:51 pm #

    Jarrett– is that a Tri-Met published map, or one you did? If the latter, there is at least one error–the #54 bus is what goes to Beaverton TC from Raleigh Hills, and the #56 goes to Washington Square.
    Interesting that Tri-Met identifies the duplex of these lines (between Raleigh Hills and Portland) as a Frequent Service corridor, even though individually each bus runs on ~30 minute headways.

  19. Jarrett at HumanTransit.org November 5, 2009 at 2:58 pm #


    Unfortunately, that's Tri-Met's map.  I'm aware of the error as well.  It's sad that nobody's bothered to fix it.

  20. Alurin November 6, 2009 at 11:36 am #

    Mizzz: Unfortunately, there are very few transit systems in the country (outside of NYC) that run past 12:30 am.

  21. Pantheon November 6, 2009 at 1:12 pm #

    To Calwatch: The kind of hedging that Trimet was involved in does not reduce financial volatility, it adds to it. Hedging a small part of expected fuel consumption could reduce volatility, by mitigating the effects of any upward or downward price movement. But Trimet hedged two thirds of its fuel output for two years. That kind of fuel hedge is just a speculation – a bullish bet on oil. It is exactly what Southwest was doing. Moreover, Trimet did its hedge in August 2008, right at the top of the market.
    TriMet’s agreement with Keybank means the following: if oil rises above the strike price of the hedge, Keybank pays Trimet. If it falls below, Trimet pays Keybank. There is nothing in that agreement that creates financial stability for TriMet, it is the very definition of speculation: windfall profits if oil rises, huge losses if it falls.
    The reason this kind of hedge adds to, rather than subtracts from financial volatility, is that oil tends to rise and fall with the economy, just as tax revenues do. So in fact, NOT hedging against oil prices is the best true hedge there is. If the economy crashes and tax revenues decline, at least your fuel costs go down with it.
    The fuel hedging program represents a big part of this picture. TriMet’s losses are $2.6 million, and the total cuts for this year are $6 million.
    Source: http://www.oregonlive.com/news/index.ssf/2009/03/post_61.html

  22. calwatch November 6, 2009 at 8:03 pm #

    Moody’s cites LACMTA’s hedging of 95% of their CNG fuel needs as a net positive: http://www.metro.net/about_us/finance/images/moodys_propC_2009-A.pdf
    Here’s how the MTA has done: http://www.metro.net/board/Items/2009/07_July/20090715F&BItem18Handout.pdf They just went ahead this spring to extend the hedging program all the way until 2015: http://www.metro.net/board/Items/2009/04_April/20090415F&BItem10.pdf It functions similarly to the Trimet program listed above (swap of funds with a third party bank or insurer) and has improved budget certainty and decreased the pain felt in energy shocks. Hedging is a tool, and is used when agencies value stability more than the alternative. It doesn’t work everywhere – for example, LACMTA’s bus riders are largely poor and so ridership tends not to fluctuate as wildly due to the economy (and is more sensitive to fare increases, like if the cost of energy spikes). The other issue is that the cost of CNG is generally more stable than oil. Still, that doesn’t mean the benefits of hedging should be discarded.

  23. Alon Levy November 6, 2009 at 10:44 pm #

    Portland doesn’t have the mode share of not just New York, but also Vancouver and Calgary, both of which are smaller metro areas.

  24. EngineerScotty November 7, 2009 at 2:07 pm #

    As discussed before–Vancouver and Calgary have the advantage of a dearth of freeways; transit is, for many in those towns (including those with autos), the most convenient way to get around.
    Unless your route involves downtown at rush hour, most trips in the Portland area can be done more easily with an automobile; as a result, Tri-Met doesn’t attract sufficient volume to realize an operational profit.
    Some urbanists in Portland do occasionally suggest closing down freeways to force people onto transit. (Here I’m not discussing the closure of obsolete routes like Harbor Drive, but of closing vital routes like I-5 or I-84). Obviously, that won’t happen–what might happen instead is resistance to any capacity improvements on existing highways.
    That said–an interesting development in Washington County is that traffic on OR217 (the freeway connecting US26 in Beaverton with I-5 east of Tigard) has gotten so bad, that officials are considering closing down most of the ramps on and off the freeway during rush hour. The highway is highly functionally obsolete–having two lanes in each direction (plus an auxiliary lane) for the majority of its length, and about 10 interchanges over its seven-mile length–and sees a lot of local as well as regional traffic. I expect that this proposal will be shot down for political reasons–but the fact that authorities are even considering such a thing says something.

  25. Alon Levy November 7, 2009 at 7:35 pm #

    Yes, Vancouver and Calgary decided to invest in transit and not overbuild freeways, whereas Portland didn’t. That’s my point – Tri-Met’s planners were building light rail, not general regional transportation of which the centerpiece was to be light rail.

  26. EngineerScotty November 7, 2009 at 11:28 pm #

    Tri-Met didn’t even exist as an agency until Portland’s freeway network was already laid out, and much of it built already. The basic freeway map was laid out in the 60s (minus a few things that didn’t get built, such as the Mt. Hood Freeway), with most of it being built by the turn of the decade. (The main exception was I-205, not completed until the 1980s).
    Tri-Met was created in 1969 to take over the bus system when the private operator went out of business. For quite a while after its formation, its charter was limited to operating bus systems–the only planning it got to do for the better part of a decade was setting up bus schedules. By the time the Portland area got around to building light rail, the damage had already been done.

  27. Alex August 25, 2011 at 8:00 am #

    Wow, ok I’m two years late with this but I was drawn in by the spambot. Just want to point out that Vancouver and Calgary both have extensive freeway networks – and I’d say you could characterize Calgary’s as overbuilt on the level of Portland’s (or almost) – they just aren’t oriented towards intracity travel and in particular CBD access to the degree that Portland’s freeway network is. Which of course impacts the competitiveness of transit for CBD commuters.

  28. Alon Levy August 25, 2011 at 2:40 pm #

    Vancouver proper has two freeways, both just on the margins of town; there’s more freeway development in the suburbs, but still much less than in an American city of comparable size. Calgary has a few grade-separated arterials, some with LRT in the median, but very few Interstate-grade freeways.

  29. Eric August 27, 2011 at 9:52 pm #

    In response to Pantheon’s comments concerning the Green Line–
    While the time advantage of rail over bus may be tiny on paper, the actual difference might be larger if the train can dependably run on time, while the bus gets further and further behind schedule as its run progresses. Given that the buses are vulnerable to so many more random delays than the trains are, I would definitely expect the trains to be more reliable. A few of the types of random bus delays that the trains are immune to include:
    – Large groups of cash payers boarding at once
    – A passenger arguing with the driver over whether a transfer is valid
    – Wheelchair loading an unloading
    – People holding up the bus to find out how to get somewhere without actually getting on
    – Traffic jams
    – Waiting for another bus to move out of the way in order to get to a bus stop
    While BRT could, in theory, avoid these types of delays, just as trains do, Portland’s buses don’t have anything that bares the slightest resemblance to BRT. The only way to avoid the above delays in the Portland transit system is to walk, bike, or drive to the nearest MAX station and ride the train from there.

  30. Alex B. August 28, 2011 at 9:28 am #

    I’d say Vancouver’s suburbs have a pretty typical freeway network for North America – just about every suburban area is touched by it. But it doesn’t matter so much because the city proper refused freeways, making transit a competitive choice for commuters.
    Calgary on the other hand, has a similar amount of freeway lane-miles per capita to Portland – highways 2, 8 and large segments of Crowchild Trail are full freeways in Calgary. The difference is that Calgary’s network isn’t oriented to the CBD to the degree that Portland’s and most US cities are.
    I of course agree that the amount of freeway per capita is big factor in mode share, but I just want to add that the network design is another factor.