Skip to main content

Neville’s contribution of rigging an electronic cyclometer and a marked rim, to count revolutions, is a significant one, happening as it did at a time when Jones Counter availability was uncertain. I believe his experiments with pressure monitoring and now with 25 meter calibration courses are interesting, but do not have as wide an application. He has said:

I have shown that there is no significant difference in accuracy between doing calibration on a 25-m rather than on a 300-m course when using rim readings in spoke intervals.

This statement has not been proven universally true. While it may be true that Neville has personally achieved comparable accuracy, it is a long way from this to a universal application. He has described an experiment done by one person and concluded that its results are universally true. Applause at this point is premature.

Pressure calibration is in its infancy. At present only Neville has data which supports his own measurements. I believe his results to be accurate – for him. Whether others will find similar results is unknown, as we have heard of nobody who has tried.

All of Neville’s methodologies require greater care by the measurer than do presently-accepted methods. The biggest impediment to general acceptance of these experimental methods is one of understanding by a reviewer – the certifier who must look at the paperwork and make sense out of it.

I have reviewed the measurements of one person who used the calibrated cyclometer and marked wheel, and it was not pleasant. Mistakes were made, and corrected. The review process took longer, as the measurer did not quite understand the process. Ultimately I certified the course.

What if someone should send me a measurement using calibrated cyclometer, marked wheel, pressure-adjusted calibration and a 25 meter calibration course? Would I then have to ask for all the supplementary information that would demonstrate that his work is OK? Neville has done this, for his own data only. It would be a nightmare to review.

Since Neville is a final signatory, this is not a problem, as nobody reviews his measurement data. He understands what he has done, and sees it to be correct.

I wonder what a measurement package sent to a certifier would look like if they used all of the new procedures seemingly proposed by Neville? Has he actually proposed them for general use, or has he simply put them out there as interesting experiments?
Original Post

Replies sorted oldest to newest

Using a shorter calibration course and rim readings by itself does not make things more difficult for the reviewer, as long as the measurer is using a Jones counter. If he's using an electronic counter, then the added difficulty arises from that, not from the calibration method.

If shorter calibration courses with rim readings were accepted, I think it should be required that the measurer report both the Jones count and the rim readings during calibration. The measurer (and the reviewer) could check that the cal constant calculated by rim readings is similar to that calculated by Jones counts, but then the one calculated with rim readings would be used since it is more accurate.

While I agree with nearly everything you say as you urge caution in accepting 25 m calibration courses even for experimental purposes let alone validations. I think you have missed one point. Perhaps you did not look at the actual numbers reported.

If you look at the numbers recorded in the trial shown in the link to the .xls file on the other thread, you will see that over the 25m course the difference between the minimum reading and maximum reading within a group of 4 calibration "rides" was approximately 4 times greater compared to the 300m course. Well, that is not surprising given the limited precision to which the spoke interval can be read (without special equipment). However, the difference between the 25m calibrations and the 300m cailbrations, which in one case was as high as 0.085% and averaged out at 0.051% (when the sign was ignored), would be the cause for considerable worry if obtained in normal practice with a Jones Counter (Model JR) on 300m - 800m calibration courses unless there was a strong wind or a steep slope.

The size of the discrepancy between the 300m calibration and the 25 m calibration - for the five comparisons done amounted to:
+0.03%, +0.05%, +0.04%, -0.06%, -0.04%.
While this set does indeed happen to average out quite close to zero. I for one would not trust any work based on such poor calibrations. The difference between the two methods for one of these five comparisons amounts to over half SCPF and the range amounts to one whole SCPF.

I teach newcomers that the SCPF needs to be preserved to cover measurement errors that are beyond the measurer's control, and not used up to cover mistakes due to slip-shod calibration procedures.

Just as important as the point I make above from looking at the figures, is the point made by several experienced measurers, that pushing your bike along a line is not representative of the actual wobble and front wheel weight when riding with the wind blowing in your face.

To see the amount of distance added by wobbles during one calibration ride I did in 1996 have a look at my article in MN no 78 July 1996, where I recorded the transverse position of the front wheel using a video camera and after analysing 2850 individual frames (2 days work!), I measured the wobbles of the front wheel to have added 0.035% to the straight line distance. A one-off experiment, never repeated, but it does give an indication of how much I wobbled.

Having said all the above, there could be some use for a modified technique over specially set up 50m calibration courses for the purpose of determining corrections for surface roughness variations between the calibration course and the race course.
I agree with nearly all your comments. All my statements go with the implication "in my hands". I try to keep my methods simple for general use in improving efficiency, but until others try them we have no means of telling if I meet my goal.
Since it takes less than 10 min to check my findings on the 25-m course while doing a standard calibration, I hope others will do so.
To be fair to your experience with the measurer trying electronic counting, he did badly mess up with even the taping of the calibration course. His problem with the Protege counter was that he could not program it properly even though it is no more difficult than doing so for a VCR. The latest counters I have found do not require programming, so this need not be a problem anymore.
Pete, I sympathize with you on the certification paperwork. It is for that reason I don't recommend the Protoge. BUT, that is not to say I don't recommend electronic counters. An electronic counter that can count backwards, as the Jones counter can, is fine. There are no gaps in numbers, or correction notes on the data sheet. I would think that the easiest way to keep the sheet clean on a 5k course is to go back to the start and ride again. It is not that big of a deal, most times.

I agree, the Jones is the simplest method to mount-and-go. The new counters appear to be very good, and I love the alignment of the digits, as they can be read more easily. I am not sure why there is a push to mark rims for more accuracy in Jones counts. If someone uses Jones readings to the half-digit, they are accurate to within about 2". Why do more work for such a minimal, and insignificant, degree of accuracy?

I also concur with Pete that pressure monitoring is for advanced measurers, those who don't mind carrying a formula with them to punch into a calculator, which then should let them know if they need to move the Start or Finish. I ride airless tires, so I won't ever use a monitor. I do have to be aware of large temperature swings, as do pneumatic riders.

Neville's 25m method, I think, has merit. If one follows the procedure, specifically to use a painted line, and weights the bike while pushing-off with one foot, one has very little wobble - much less than when one is riding while pedaling. Neville's results seem to yield the same net distance for a 5k, as a 300m calibration course. Again, I think the key is to use a straight, painted line, and stay within the 4" width. Although, if a rider went outside the line, it would yield a longer race course. Minimally longer, so I don't believe that is any worse than someone calibrating with a stiff cross-wind and having more wobble than normal (which would yield a longer race course than one calibrated with no cross-wind).

I think marking a rim more-often than every .05 revolution is wasted effort. Since the .05 markings are only about 4" apart, it is not difficult to estimate the intermediate hundredths. I am fortunate that my rim has 20 spokes, so I only marked every .25 revolution. Simple to estimate hundredths.

If Neville is correct, that the measurer had other errors, then that measurer needs to spend a little more time learning the methodology for electronic measuring. I would no sooner discard the electronic method due to his errors, than I would discard the Jones as "too complex" if someone didn't roll backwards if they overshot, or had problems identifying the Shortest Possible Route. Everything takes time to learn, even our, in our minds, simple measurement methods.

I will back Neville that the electronic counters work, and work well. As a Totalizer is available for $25, and the wiring for it is cheap (my source didn't charge me at all), his two-Totalizer method is viable, and is a reliable way to see if one counter had errors. The only way the counters would err is if the measurer stopped with the sensor aligned with the receiver. Easy to avoid. Two counters overcome that issue completely.

My 2-cents regarding electronic counters, and 25m cal courses.
the method requires no skill

Neville, let's show some respect! I think Pete has plenty of skill, and is fully capable of your 25m method Big Grin

Pete, I don't think Neville was saying "one rides the same on 25 meters...". I think he said that the calibration factor is demonstrably the same when the 25 meters is ridden exactly as he describes, as compared to a normal 300m cal ride.

By the way, if anyone is going to the RRCA convention in Cincinnati next weekend, I would love to meet with you. I will be there.
I think marking a rim more-often than every .05 revolution is wasted effort. Since the .05 markings are only about 4" apart, it is not difficult to estimate the intermediate hundredths. I am fortunate that my rim has 20 spokes, so I only marked every .25 revolution. Simple to estimate hundredths.


If we assume the Jones can be accurately read to one half count, for a 300m cal course that translates to a reading accuracy of about 20% of the SCPF. To achieve that same level of accuracy with a 25m cal course, you would need to be able to read revolutions to the nearest 1/500th of a revolution.

One could say that we don't really need to be that accurate, but I think we would all agree that we need to be able to read to within 50% of the SCPF (even though this would be a huge degradation of the current accuracy with a Jones and a 300m cal course). This requires the ability to read to the nearest 1/200th of a revolution on a 25m course. I don't think it's possible to do that accurately with only 20 marks on your rim.
Nev says:
Obviously when a measurer demonstrates he gets the same result by both methods he can begin to apply the new method on its own.

Not if he expects to get his course certified he can't, unless a full-length calibration set is submitted with the measurement data.

Your examples of a highly-skilled and technically adept person trying out new approaches are interesting, but only that. The fact that you have what you believe to be revolutionary results doesn't begin to justify turning present procedures upside-down.

I see what you are saying, now. But, if a full click of a Jones is about 3.5" to 4", reading to a half-click on a 300m course is considered sufficient.

However, I don't see a way to accurately estimate it to much less than a quarter-click. Even if a rim is marked in 100ths, we are still guesstimating exactly where the next click of the Jones is starting, aren't we?

For that reason alone, the 25m cal course may be too short.
This topic is interesting, but all it has been is food for thought. Nev's method of a 25 meter Calibration Course may work for him and others, but it is not the standard for the RRTC used in measuremnt.

I hope he was not serious when he states that a measurer can use this method. This is not the case!

Neville, thanks for the idea.
However, I don't see a way to accurately estimate it to much less than a quarter-click. Even if a rim is marked in 100ths, we are still guesstimating exactly where the next click of the Jones is starting, aren't we?


I'm not sure I follow you. If you are using rim reading to do the calibration, the Jones counter is not part of the process at all. When I suggested that you would want to report Jones readings as well, that was only to reassure the measurer and the reviewer that no gross calculation error was made when determining the cal constant from rim readings.

I have done a lot of calibration rides with rim readings. Based on my experience with a rim marked in 100ths, I believe I can tell revolutions down to 1/200th. More precise than that is very difficult no matter how many marks you have on your rim, because the marks on the rim do not actually touch the road surface. Slight changes in your viewing angle of the rim marks and the road marks can easily change a reading a couple 1000ths or so, and that's 20% of the SCPF on a 25m course.
Ha ha well done Mark. My understanding of the reason for the imprecision of the rim reading method -- ie distance of rim marks from the road surface is exactly what you have expressed.

I think to get a 7 foot calibration course to work we would need get the rim markings to be almost in contact with the nails at the ends of the calibration course. I have got some ideas about this and I have a small group monitoring my work and contributing ideas. Please email me if you would like to contribute to the design process. I am not doing the design on this BB as the BB format is not particularly convenient.

My confusion may stem from my lack of a clear understanding how one can do a calibration with the Jones not being part of the process.

If I have a 300m cal course, I begin at one end, with my rim's zero mark on the nail. I arrive at the other end, and read my rim mark at the end nail. It is .23 of a revolution. So what?

I must assume that you know from previous rides that you will have 4123 full Jones counts between the nails, so you add that to the .23 reading you now have. Is that correct?

If so, I still feel it is a bit of faith to believe your Jones is working properly, do your cal without noting the Jones counts, then going to the race course. How will you know if your Jones is malfunctioning?

I ask, as I discovered my Jones was malfunctioning on a cal ride. I knew how many clicks I normally had, but had hundreds more on a ride. The next ride was also hundreds off, so I knew my Jones was trashed. Better to find out on a cal course, than to measure a race with a faulty Jones, and still not know it, since you don't look at the Jones on a cal course.

Am I following this correctly?
My confusion may stem from my lack of a clear understanding how one can do a calibration with the Jones not being part of the process.


From previous calibrations that I have done with my Jones, I know that my cal constant is about 11790 counts/km. Next rim reading calibration I do with a 100m cal course, the only data I really need to collect is the last partial revolution, say I was at 0 at the start of the cal course and 44 at the end. So the wheel rotated x.44 times. The correct x (number of whole revolutions) is the one that gives a cal constant close to 11790. Any incorrect x will give a constant that is more than 200/km away from 11790, and constants just don't change that much.

But in real life, just to make sure, I always make a note of the Jones counts on my first cal ride so I have a same-day sanity check. But that Jones count reading doesn't figure into the calculation I do to determine the cal constant from my rim readings. It only provides a baseline constant that I know my rim reading constant should be close to. If it's not, I know something must be wrong, and in rare circumstances I suppose that something could be the Jones counter itself, as you suggested.

I should note that all of this only applies to measuring I do that is not for USATF certification. For USATF certifications I, of course, follow the 300m requirement.
Failure of the Jones should be even rarer in the future if advantage is taken of the feature that allows the meter to be disconected while not measuring. Obviously though, you can check it by taking readings on one of the calibration runs; you just would not have to take readings on all eight runs. Better still you could check it over much longer distances on the actual race course against an electronic counter.
Mike and Mark give theoretical reasons as to why the 25-m calibration will not work, but ignore the fact that it has worked perfectly for the many times I have tried it: in a real-life situation it gave a calibration factor that deviated from that derived from a 400-m course by only negative 0.008%. I am reminded of the fact that eminent physicists once proved that it was impossible for man to fly!
I much too busy to reply at length with more explanation of my analysis of the published data on 25 m cal courses.

I am watching the araldite set on my new 7 foot calibration course. That is going to take another 16 hours of watching. Once it is set I will have to take quite a bit of data to evaluate performance. Then there will be the analysis to do, so I dont expect be able to publish results for at least another 24 hours.

My goal is a modest 0.03% accuracy for bike "pushes" not for "rides", but most likely I will fall short of that and the 7 foot calibration course may yet be consigned to the rubbish tip! If not, then I will be deploying the course (it is portable) on a variety of rough surfaces such as grass and gravel which I have never properly characterised before. This will be useful in checking our assumptions for the amount of off road surface we allow in the UK see HOW MUCH OFF-ROAD SURFACE IS PERMITTED BEFORE A RACE IS MULTI-TERRAIN?.
in a real-life situation it gave a calibration factor that deviated from that derived from a 400-m course by only negative 0.008%.


In the data you showed, it deviated by 0.03% to 0.06%, or 30% to 60% of the SCPF, correct? I didn't see your example where it deviated by only 0.008%.

I gave a practical reason why I think the accuracy of this technique might be limited. I don't think you can take measurements with precision less than say, 30% of the SCPF. But with a Jones counter and a 300m course I think your precision is only about 20% of the SCPF, so not a big difference. I was mostly making the point that I think you will need to have a lot of accurate marks on your rim to make the technique work.

I was not trying to say it's impossible for the technique to work. I'm still planning to try it.
But with a Jones counter and a 300m course I think your precision is only about 20% of the SCPF.

I think you are spot on with this assessment of the Jones counter precision. I recently calculated virtually the same maximum error for a single ride, assuming the Jones read to the nearest half digit, and I also allowed 5 mm at each end for alignment error with the centre of the nail. I reckon I can get well with in the head of the nail which is 11 mm diameter.

I notice that two IAAF grade A measurers always clamp their wheel between each of the 4 rides, and that takes the overall precision down to well less than a tenth of the SCPF. However, I would not think their overall result is that good, because they visibly wobble slightly as they ride (not just during the pushoff). I am afraid these grade A measurers are never going to get representative results from pushing their bikes along a short cal course unless they can keep control of those wobbles and somehow allow for them.

Actually I think that the variation of wobble amount contributes to the scatter I have observed in a set of cal rides taken with a solid tyres under well controlled conditions on a 650.60 m cal course which I recorded in my article in MN no 78 July 1996. N.B. these data were taken reading the Jones to 0.1 of a digit - Some optimistic interpolation there perhaps - but it is a bit more accurate than those who read to nearest half of a digit.

Add Reply

Link copied to your clipboard.