Response to London Cycling Design Standards Consultation

Below is our response to the consultation on the new London Cycling Design Standards, which closes today.

Our comments are on a chapter-by-chapter basis, but our principal overall concerns are

  • a lack of detail on good cycling design, and too much inclusion of dated, substandard techniques;
  • an absence of an objective approach to cycling infrastructure, particularly the discarding of a matrix-based speed/flow approach to cycling provision, in favour of a 'place/movement' approach that is open to interpretation;
  • the inclusion of techniques that rely on (often counterintuitive) individual behaviour to ensure safety, rather than employing sustainable safety techniques that ensure safety flows from the design, and is forgiving of human error;
  • and 'shoehorning' of cycling design into existing street layouts and types, without proper consideration of whether they are genuinely inclusive.

We would like to see a Cycling Design Standards more tightly focused on the quality of environment that people cycling experience, based around the separation criteria employed by the CROW manual, which ensures that cycling is a comfortable and attractive experience, for all, with minimal interactions with motor traffic.

These criteria should be backed up by employment of Dutch sustainable safety principles in all aspects of the way cycling is designed for; to create recognisable road and street types; to minimise risk; to respond to natural human behaviour; and to create pleasant and safe cycling conditions.


[Note: the sheer size of this document - around 350 pages - means that we cannot possibly pass comment on every single aspect of it, so omission to mention something should not be taken as implicit endorsement]

Chapter 1 - Design Requirements

This chapter sees an early mention of “high quality infrastructure” in the second sentence (page 2). This is a decent start. Yet the document then surprisingly advocates wide bus/cycle lanes as a solution (page 9) when this has been largely discounted (or never introduced) in countries with high cycling rates. The PRESTO series also advises against shared lanes under most circumstances, and describes bus / cycle lanes as a transitional solution, at best. 

A statement lacking confidence is found on page 10 – discussing separation methods for cycling. The claim that “some busy, narrow roads can never be truly made safe for cyclists” seems quite dubious. As the decision to create a true cycle network is largely taken by meeting ‘core design outcomes’ in conjunction with traffic volume, speed, and vehicle type, a negative statement like this is not necessary and should be removed.

The ‘nine street type’ methodology taken from the Road Task Force and outlined on page 13 is a sound concept, but the breakdown is perhaps still quite a bit too general and open to interpretation. Nevertheless this may provides a framework for discussion about the purpose of a road: is it a place, or movement, or somewhere in between - and how should we be moving it into either of the two place or movement categories.

There is a hint at a two-tier network in the description of the Quietways (that are to complement the Superhighways). The Quietways, the draft states, “are principally not aimed at existing fast, confident cyclists”. This could easily be read as a patronising “Boris-type” statement favouring the fit and the brave (and Quietways should really aimed at all types of 'cyclist', in any case) and the document could be more inclusive if it were removed.

The thorny issue over a local authority’s duty to secure the “expeditious movement of traffic” is dealt with on page 21 where ‘traffic’ is “explicitly defined as including pedestrians, cyclists and motorised vehicles”. A wider resolution of this hang-up from the 60s Golden Age of Motoring is not sought, thereby leaving it in the road safety and cycling campaigners’ hands to deal with.

The first chapter ends with listing the Disability Discrimination Act and the Equality Act. Finding these bits of legislation that support inclusivity in the text must be a good sign for taking responsibilities seriously to provide adequate facilities.

Chapter 2 - Tools and Techniques

More processes are introduced in this chapter. The delivery steps of the cycle infrastructure are sensibly outlined on page 25. Confusingly three Mini-Hollands are then described in addition to the Superhighways and Quietways. How these discrete areas meaningfully feed into the overall network creation (other than offering three temporary test beds) is unclear.

Then the Cycling Level of Service (CLoS) procedure is described. It is an audit system and links to the six outcomes: safety, directness, comfort, coherence, attractiveness, adaptability - this last being the addition to the list otherwise taken from the Dutch CROW manual. As we will see with later chapters, 'adaptability' - while useful in its own right - is not strictly a measure of 'service', and should not be included here.

A full matrix is then set out, further breaking down the outcomes on pages 31 and 32. This matrix really warrants a consultation in its own right.

Junction assessment is another interesting process (page 33) that the document introduces. The procedure outlined looks worthwhile and offers another decision-making tool by visualising, quantifying and putting something on the table for discussion. Various other methods then follow quickly one by one: mesh density analysis, accessibility classification, area porosity analysis, CLoS audit, ending by giving an example. On page 49 it does appear that vision communication and early community engagement may have been missed out from the delivery process that entirely focuses on engineering project stages.

Chapter 3 - Cycle Lanes and Tracks

There is some good detail in this chapter, but unfortunately it is quite woolly on what kind of provision should be applied under particular conditions, and there are a few strange recommendations that don't correspond with best practice, which isn't set out clearly enough.

The Cycling level of Service (p.59) is a good approach, based it seems on the CROW manual characteristics of safety, directness, comfort and attractiveness. There are useful examples of how these criteria should be met, and how they might be failed. However, in what appears to be a nod to light segregation, 'adaptability' is added as a characteristic, which seems strangely irrelevant - 'adaptability' is not important to those actually cycling along a road or a street. The key criteria is for the cycle provision to be good; 'adaptability' has no bearing on the level of service that's received. There is a place for adaptability to be considered; it just shouldn't be here.

There's nice language about the importance of inclusivity, and of catering for anyone who wants to ride a bike, and dimensions for non-standard bicycles, and the importance of avoiding barriers, and so on. Effective width is covered clearly, with recommendations about how to maximise it, through the use of forgiving kerbing.

There's also a clear statement on how cycle tracks can be designed properly at junctions in paragraph 3.5.11 - maintain the track at the same level by use of a raised table and apply corner radii that are as tight as possible,  followed by a useful discussion on how to minimise, reduce, or even avoid entirely conflicts at these junctions. These are important points and should form a central part of the way cycle tracks are designed.

Equally, on the subject of making roads one-way, there is good detail on how to create more space for cycle provision by converting roads to one-way for motor traffic, with a good series of visualisations on how to achieve this. And two-way cycling on one-way streets is covered nicely - this really should be standard practice now.

But unfortunately these good recommendations - of which there are many - are lost in a long chapter which contains an awful lot of things that shouldn't really be in a design manual at all, at least not without explicitly clarifying that they are interim measures, or that they should only be tolerated under limited circumstances.

Perhaps the most troubling general issue is the way this guidance explicitly abandons a matrix-based approach to cycling provision - one that suggests what should be appropriate, given the volume and speed of motor traffic on a particular road or street. Instead there is a rather woolly 'place and movement' approach that makes vague suggestions based on the place or movement characteristic of the road/street in question.

This runs counter to the way the Dutch design for cycling - explicitly matrix-based - as well as London Cycling Campaign policy recommendations. It opens the door to cycling provision being substandard on busy roads that are deemed to have more of a 'place' function. Motor traffic speed and volume would provide a more objective standard of whether a road or street is a genuine place, but objectivity appears to be replaced by more subjective guidance, which is worrying.

To take just a couple of examples of how this approach might not work, from this chapter itself. It states that dedicated cycle lanes should not generally be used for streets with volumes above 500 motor vehicles per peak hour without a 20mph limit' which is fine, but with a 20mph limit in place, how much motor traffic can be tolerated with simple cycle lanes? There is no answer in this chapter. A matrix would provide much-needed clarity.

Likewise, while the chapter is generally good on cycle streets, with guidance on how they should employed appropriately, it argues that they should be used only where motor traffic volumes and speeds are already very low. But how low? Again, no clear answer - this is all too subjective. 

More generally, throughout the chapter, substandard provision is included, without comment. There is a long section on the apparent advantages of bus lanes - true as far as it goes - but no statement that they are not truly suitable as cycling provision, and appropriate only as an interim measure.

There's a statement that shared use footways should have a maximum design speed of 10mph - no assessment of whether the kind of provision that necessitates this low design speed should even be contemplated in the first place.

Light segregation - along with mandatory cycle lanes - are correctly identified as cheap and easy ways of transitioning to more permanent measures. However, there is an implication that light segregation has benefits of a more permanent kind, by not 'constraining' cyclists in the way that physical segregation does. But if people are being 'constrained', that suggests the provision (be it 'light' or 'hard') is inappropriately narrow - employing light segregation should not be a way of ducking these issues.

There are also curiosities - why should armadillos be placed on the inside of cycle lane markings? Why is it true that light segregation makes it easier for pedestrians to cross the road than properly-designed cycle tracks, with forgiving kerbing?

Junction priority also provokes some worrying suggestions, particularly the statement that cycle tracks

must be broken and converted into [cycle] lanes at priority junctions in order to reintegrate cyclists briefly with general traffic, enhancing their visibility.

This 'must' is directly contradicted by the recommendation in paragraph 3.5.11 quoted earlier (which comes just a page later in the LCDS!) which suggests cycle tracks can be continued through junctions. Indeed, maintaining a cycle track through a junction, at a raised level, with continuity, is both safer and more attractive than attempting 'reintegration', which is not employed by the Dutch, or by the successful Old Shoreham Road scheme in Brighton, which maintains cycle track continuity.

Indeed, this chapter even curiously recommends bending in a cycle track, to the carriageway, in addition to the standard Dutch practice of bending the cycle track away from the carriageway at a junction, where space is available.

There is also a suggestion that - to achieve this 'reintegration', the physical protection must be broken some distance before the junction - 20m or more before the junction on roads with a 30mph limit - a recommendation which removes the ability to constrain turning vehicle speeds with tight geometry at the corner, and encourages left hooks (to say nothing of the attractiveness of 'reintregrating' with 30mph traffic). This is a design flaw with the current Superhighway 2 Extension.

Physical protection should, instead, extend right up to the junction, to keep vehicle speeds low. Equally, changing from 'lanes' to 'tracks' and back again means the height of a cycleway is constantly changing; not pleasant for cycling. A cycleway should be continuous, smooth and safe, and the physical design of the street should reduce or eliminate the danger of turning conflicts. It seems here that, instead, the quality, comfort and safety of the cycle provision is being sacrificed in an attempt to achieve the same objective. 

There is also something of a reliance on designing for the 'primary position'. If people have to adopt this position to keep themselves safe, then something is going wrong with the design of the street. For instance, on page 64, it is recommended that a primary position should be designed for, to deal with 'high left-turning flows'. Well, that should be dealt with properly, either through separating movements, or by reducing those high vehicle flows - the primary position shouldn't be endorsed as a compensating strategy. 

Similarly, lane widths of below 3.2m - which necessitate the primary position - should not generally be recommended, except in low traffic areas. But this isn't clear from this chapter, which simply says avoid lane widths of 3.2-3.9m.

This Chapter of LCDS could be condensed, with poor, weak or subjective recommendations stripped out, leaving the good recommendations. That would be a considerable improvement, along with the retention of a matrix-based approach that we are not sure it is sensible to discard.

Chapter 4 - Junctions and Crossings

As with Chapter 3, on Cycle Lanes and Tracks, the good details on junctions that are contained here are lost in a long chapter that contains a lot of sub-par infrastructure, and with little technical information on how to implement the better recommendations.

And also - as with the previous chapter - an objective, statistics-based approach to what kind of treatment is appropriate (according to motor traffic volume and speed) is absent, replaced by a subjective approach that uses words like 'high' and 'low' without defining what they mean.

For instance, the chapter recommends removing priority markings on page 146 -

In some cases no road markings may be considered to be necessary where vehicle speeds and flows are low

Well, how low? The Dutch have clear guidance on when this kind of treatment - which employs a 'priority to the right' approach, without road markings, is appropriate. But 'low' is too vague. 'Normal roundabouts' are similarly described as suitable for cycling in 'low flow' conditions (4.5.4) - but again, how low is low?

Likewise, 'shared' toucan crossings are mentioned on p.137, before the chapter states that ‘high demand’ should merit a ‘parallel crossing’ (i.e. separation of pedestrian and cycle movements). But what is 'high demand'? It's not quantified. This is worrying, because it opens the door to compromised infrastructure.

Figure 4.8 on p.148 has some useful detail on what lane widths are acceptable (at refuge islands) but instead of a clear PCU-based approach, giving motor traffic volume, the headings in the table are instead worded 'traffic calmed' and 'no traffic calming', which is not at all clear.

There's also a reliance, on the one hand, on 'bodges' of a pedestrian kind, and inadequate infrastructure suitable for dealing with movements at busier junctions, on the other. At the low end, a straightforward zebra is described as a 'cycle crossing' in Figure 46 on p.137, which really isn't acceptable - a zebra crossing is for pedestrians. Zebra crossings, in addition to shared use pavements, are described as a 'pragmatic choice' on p.141, but should this kind of provision be endorsed, at all?

And on p.144 -

consideration could be given to allowing cycle left and right turns by diverting cyclists onto shared areas of footway and parallel or toucan crossings

Left turns should be achieved with cycle-specific infrastructure, not by pushing pedestrians and cyclists into conflict with shared use footway. 

The impresssion given is that cycling has to be 'fitted in' around existing infrastructure, and not prioritised. Crossings should be placed where people want to cross,  for directness and comfort, as in paragraph 4.2.1.

At the other extreme, there is far too much information on Advanced Stop Lines - a full eight pages of various forms and types. This is a form of provision that the chapter itself admits allows comfortable right turns 'only if the cyclist arrives during a red phase' - i.e. infrastructure that only functions on a part-time basis, at best.

ASLs really aren't adequate, and should only be employed in genuinely low-traffic environments, if at all. 'Head-start' signals - given a mention here - do not address the fundamental problems with ASLs. 

As with Chapter 3, cycle-specific provision is recommended against in some locations, on the grounds that it might encourage a 'secondary riding position' (p.155). And roundabouts that 'make it easy to adopt the primary riding position' (p.180) are endorsed, without reflection on whether this style of riding is attractive. Again - the street or junction should be designed properly, so that strong riding positions do not have to be relied upon. The primary position should not be a compensating strategy for inadequate design.

It is also disappointing to see the 'always stop' form of signalising cycling movements (as employed at Bow Roundabout) being endorsed, described here as a 'cycle gate'. Paragraphs 4.4.21-24 contain the phrase 'two stop lines for cyclists’, which is telling - really, this is a bodge used to retain motor traffic capacity, at the expense of convenience for people cycling, and it flies in the face of the points made in this same chapter about signalling for cycling needing to be 'equitable' and 'convenient'.

Section 4.4.34 talks of 'support for right turns', making the good point that these need be achievable in a comfortable way. But it has a strange way of presenting the case for a two-stage right turn -

When faced with a difficult right turn, many cyclists choose to make the turn in two stages on carriageway.

Surely the word ‘choose’ here should be replaced with the words ‘are forced’? Nobody chooses to add inconvenience to their journey, and there are signs here that the 'two-stage' right turn is really being endorsed as an easy compromise, rather than being developed properly for the convenience and comfort of people cycling.

While there is some tentative discussion of using Dutch signalised junction design, incorporating 'protecting islands', and some detail on separately signalling left turn motor vehicle movements from cycling (although not explained, or laid out, clearly), there isn't anything at all on the potential for 'simultaneous green' signalling for bicycle traffic, which would remove conflict with motor traffic at signalised junctions completely.

There needs to be much more detail on both of these essential techniques for reducing conflicts at signalised junctions.

Nor is there wholehearted endorsement of the concept of removing left-turning bicycle traffic (or bicycle traffic progressing straight ahead on a T-junction) from signal control; it is only stated that this may be appropriate 'in some instances', when really this should be a default to aim for in all circumstances.

There is nothing - at all - on cycle track treatments at priority junctions, which is really quite alarming - the only diagram of a continuous footway across the junction mouth is marked up with a cycle lane (p.153).

Meanwhile there is an over-reliance on painted markings at major junctions - for instance annular painted cycle lanes around the perimeter of roundabouts (really quite bad design) get a mention, as does the concept of

continuing the ahead cycle lane past the left-turn slip lane [which] will require left-turning vehicles to cross the cycling facility

Slip lanes and painted cycle lanes really shouldn't be mixed at all; paths will be crossed at shallow angles, and that's not a good idea, at all. Further, the notion that people should have to 'make an early and clear lane choice' fits uncomfortably with the intentions to create inclusive cycling conditions - these kinds of lane changes are unattractive, and difficult, for most people.

Also concerning is Diagram 4/04 on p.147, which appears to 'resolve' a pinch point problem with a painted advisory cycle lane that will be overrun by motor vehicles. In these locations physical protection is required, rather than advisory markings, or a design that does not involve conflict. Figure 4.8 seems to formalise 'pinch point' widths, and talks of 'alternatives' to them, rather than recommending against them altogether

The chapter is a little weak on roundabouts generally, but this is unfortunately mainly a function of the fact that good design (proven elsewhere) is still being trialled, and awaiting approval - and a historical failure in Britain to design properly for cycling at larger roundabouts. (The word 'accidents' used here on p.178 should be replaced with 'collisions').

It is interesting to see this chapter recommending against the use of mini-roundabouts anywhere on a cycling route, raising good points about how poor they are for comfort, particularly with obtuse junction angles.

Low-level signals - recently given trial approval by the DfT- appear to be being endorsed here as a stand-alone signal, without any accompanying full-size high level signal heads. This is a mistake. Low-level signals are simply an extra bit of convenience; they should always complement full-size high-level signals, particularly as low-level signals are very difficult to see from a distance. There are safety and convenience reasons for always marrying the two.

Finally, it is pleasing to see this chapter stating that ‘gyratory removal should not be an end in itself’, arguing that

The focus of any gyratory redesign should be on enabling more direct journeys with less delay, particularly for pedestrians and cyclists, and on allowing more ‘conventional’ approaches to be taken to cycling provision and to management of motor traffic speed and volume. This may only entail part-removal or partial remodelling of a gyratory or one-way system.

Chapter 5 - Cycle-friendly street design

As with earlier chapters, 'adaptability' features again as a quality criteria, which we feel is a mistake. While it has value in its own right, it is not 'good design' for cycling. Someone cycling along a street or road will not care whether the design is 'adaptable'. Likewise 'cheaper and flexible' (p.207, while worth considering in some circunstances,  are not good qualities for someone experiencing an environment; it's perfectly possible to design high quality permanent cycling infrastructure that is not visually intrusive.

Indeed, a theme in this chapter appears to be fitting in cycling around existing street design, rather than focusing on what makes a good cycling experience, and designing appropriately. 'Lane narrowings' feature on page 199. While these are useful in their own right, they are not attractive as cycling provision. They necessitate taking a 'primary position' - uncomfortable for most people - to prevent dangerous overtaking, and also make it very difficult to filter past stationary traffic. If lane narrowings are to employed, it should be in a motor traffic-specific context - that is, with separate cycling provision alongside.

Likewise in paragraph 5.2.3 on p.194, there is this slightly back-to-front formulation -

Street types that are more likely to be amenable to targeted traffic volume reduction and cycle permeability measures are those with lower movement functions and higher place functions 

Quality design for cycling should come from a network approach, with the routes decided upon first, then appropriate treatments employed on that network to ensure good standards across it. The approach here in this chapter, however, suggests that the cycling network should fit around the existing 'low traffic' streets ('low' again not defined, as in previous chapters), which is problematic, because that may not be (and probably won't be) the ideal network.

On page 208 we are told that 'it is advisable to start [designing] with decluttering in mind.’ Why? Surely the starting point should be designing well for cycling and walking - if that involves some decluttering, then that should be employed, but the starting point should be the outcomes for the user, not ‘decluttering’. In this context, Byng Place in Camden appears to be being cited as an example of good 'decluttering' practice, but this is a location where the cycling (and walking) environment has actually been worsened thanks to 'decluttering'.

With 'shared space' more generally, this chapter mentions how it requires 'all users negotiate their right of way cooperatively’. But this is not inclusive design for cycling, especially in environments with higher traffic volumes. Not everyone wants to ‘negotiate’ with motor traffic, nor should we be designing environments that force them to - this is even acknowledged in 5.5.11. So why is this being recommended? Again, this chapter is crying out for clear PCU/day limits for when this kind of ‘sharing’ is appropriate.

Indeed, there are other measures mentioned here in this chapter that are only appropriate for low traffic environments - for instance, 'homezone' treatments (p.198), and the removal of centre lines and addition of cycle lanes on narrow roads (p.201). Yet there is no specific indication of the circumstances under which these treatments are appropriate - certainly not on roads or streets carrying above 2-3000 PCU/day. But again objective criteria are missing.  

Speed cushions (p.204) should not be recommended, at all, particularly because they cause conflict between motorists and people cycling heading for the gaps between them. Full-width sinusoidal humps should be the standard treatment, if humps are required, with bypasses for cycling, if appropriate.

(There is also worrying talk - as with earlier chapters - of 'returning cyclists to the carriageway at junctions and side roads', on p.207, which is something that should not be employed;  cycle tracks should be consistent and clear through a given junction, with design ensuring safety, rather than giving up).

It is heartening to read this in a cycling design manual (p.212) -

International best practice shows that cities with good quality, joined-up cycling networks do not generally rely on footways shared between pedestrians and cyclists in inner urban areas. That is not to say that shared facilities might not have their place in certain circumstances, particularly alongside major arterial roads, but to stress that they are an option offering a level of provision that ought to be explored only when options that provide separated space have been exhausted.

along with an acknowledgement that space can be taken from footways, where appropriate, and without impinging on pedestrian comfort.

However, the entirety of the section on ‘shared use’ (p.212-24) could be condensed considerably, and made more easily comprehensible with the simple addition of a table indicating under what conditions sharing between people and walking, based on pedestrian volume/per width/per hour. We suggest a straightforward borrowing of the CROW manual recommendations here.

Worryingly it seems that ‘sharing with buses’ is being endorsed as ‘cycling provision’ (page 225); this is not acceptable. This makes discussion of ‘on-carriageway’ treatments around buses rather moot, because this should not be being designed for in the first place! Bus stop bypasses - briefly mentioned here - need much more development, as these are the only acceptable way of designing for cycling along bus routes. Unfortunately Diagram 5/04 (of a bus stop bypass) is unhelpful, because it suggests that the bypass is an ‘option’ for people to cycle through when the bus is stopped, alongside parallel on-carriageway provision, which isn't the function of a bypass, at all. We need coherent design, based on continental best practice.

With regard to kerbside activity, it is a bit meaningless to argue that (p.236)

segregated cycle lanes/tracks will generally preclude all loading activity, unless it takes place in marked bays on the offside of the cycle tracks and involves goods that can be delivered across the tracks.

The qualification shows that cycle tracks don't actually preclude loading activity, at all, any more than pavements preclude it. If goods can be loaded across a pavement, then they can be loaded across a cycle track and a footway just as easily, provided the cycle track is designed well.

Finally, there is a poor layout in Diagram 5/05 on p.239 - cycle lanes should pass on the inside of parking bays, rather than running on the outside, which is more unattractive, as well as more dangerous. This poor design is repeated, again, in an illustrative photograph on p.241. Likewise, rather than running cycle lanes outside taxi ranks (similarly unattractive) there is no reason why taxi ranks cannot operate in the same way as bus stop bypasses, with a waiting island, and a cycle track passing inside the taxi rank.