Skip to main content

If this is an old topic, forgive me and please point me to the archived discussion if you know where it resides.

We use the certification measurement that results in the shorter of two measurements, assuming both rides are good. Then we adjust as necessary to get to the exact stated course distance. When we use the "Sum of Shortest Segments" to measure a marathon, or any other race, we select the lowest-count ride of each section because we assume it is the most accurate measurement of that section.

When we calibrate, we employ an opposite methodology. Is this just because creating a longer cal constant tends to defend against short course measurements?

If I take 4 rides on a cal course, and one ride produces a number that is a count lower than the other three, wouldn't this likely be the most accurate ride? If so, using only the lowest count will produce a slightly lower working constant. Yet wouldn't doing this tend to produce a more accurate course?

We are working to define an accurate course, not one that is simply guaranteed to be at least the stated distance. What is the limit for a course being too long? For record purposes, I understand long is OK, and short is never OK. But, how much longer is still OK?

It seems to me that the lowest cal count would likely be the most accurate ride on a relatively straight and flat cal course. Is our justification for using a constant that possibly incorporates less accurate cal rides a rationalization that our course riding will be less accurate than our calibration riding?
Original Post

Replies sorted oldest to newest

Seems that four cal rides would yield counts that mimicked starting and stopping as we measured the race course. While one ride may be fewer counts than the rest, all of them taken together should yield a more-accurate measurement.

If we only used the cal ride with the fewest counts, each split may be 15,000 clicks, instead of 15,015 if we used the average of 4 cal rides. This removes all of the SCPF. If we wobbled more on the race course than the cal course, but used a "perfect" cal course ride to calculate our split/overall length counts, we could easily come up short on our race course ride.

I think the way we do it is a very safe way to get courses that are at least as long as the stated distance. I would venture a guess that extremely few courses, if any, are actually .01% longer than advertised distance, if anyone was able to measure absolutely accurately with steel tape.

Just my thoughts.
Four precalibration calibration rides were required by Ted Corbitt in the days before we had an RRTC. Only two postcalibration rides were called for, and the precalibration was the one that was used. As our methodology grew, the four rides were not seriously questioned, but when larger constant was adopted it was seen that four rides should be taken on both precal and postcal.

Ted Corbitt’s work was a continuation of the British work done by John Jewell, who investigated many different methods of measurement. His 1961 monograph on the subject “Notes on the Measurement of Roads for Athletic Events” may be seen by going to www.rrtc.net and clicking on http://www.runscore.com/coursemeasurement/ Early Corbitt work may also be seen at this site.

Jewell’s comprehensive work did not specify any particular number of rides.

As Bob Baumel's idea of larger constant grew, it was seen that it made no sense to have only two postcal rides, especially if the postcal turned out to be the larger. So, 4 precal and 4 postcal became the standard.

In 1982, in Measurement News #2, Bob Letson began looking at the idea of shorter calibration courses. As RRTC Chairman I supported the idea. The next several issues of MN followed the investigation of the feasibility of change, and within a year it became the new standard. Previous to that time the minimum length for a calibration course was 1/2 mile or 1 km. As this takes considerable time to lay out, as well as to find a straight road of the proper length, few new calibration courses existed. Adoption of the present 1000 feet or 300 meters had a positive effect on the creation of new calibration courses. It also greatly reduced the calibration riding required.

As for safety against short courses, it was never considered that the largest single calibration ride should be the one used. It was felt that the higher 4-ride average, coupled with the short course prevention factor, provided adequate protection against short course creation.

I may have a few facts wrong here, but this is how I recall it.
Last edited by peteriegel
Pete, thanks for the historical info. I enjoy learning about the genesis of our methodologies.

I enjoyed reading your 1982 letter to the RRTC. Though RRTC did not adopt your triangle calibration protocol, you did a good job explaining the usual differences we can expect between calibration rides and measurement rides.

Having laid out one 1/2 mile calibration course in blazing hot July sun along a busy road with speeding cars mere feet away, I for one was grateful for the 1000 foot/300 meter standard you helped establish.

You and Duane have explained the answer to my question well. This corroborates my thinking, and it makes sense. So, refining this a bit more, let me put forth a follow-up question: Two of the cal courses I use most commonly have no easily track-able straight line that one could use to follow the cal course setup path. I have to ride very carefully on these courses to ensure that I do not get a calibration count variation of more than 2. A cal course I use in Leesburg, VA was thoughtfully laid out by Bob Thurston 2" to the side of a perfectly straight concrete/asphalt dividing line. I find it much easier to get all my rides right on or within a single count on this cal course.

My point is that on the cal courses I use most frequently, I am sighting the calibration line in somewhat the same way as I am sighting the tangents on road measurements. Therefore, I think I can make a semi-cogent argument for using the shortest count on my cal rides on these courses, which is why I brought this up. I do not think I usually wobble more on a course measurement than on a cal ride. Yet I easily see where this is more likely to occur on a measurement ride. I have no impetus to change the 4-ride-average protocol. But this discussion helps me grasp how and why we do things as we do.

Thanks.
Last edited by pastmember
There are other reasons to take the average beyond the question of how straight or crooked you ride. For one, tire temperatures may take some time to stabilize. Several times I've gotten significantly fewer counts on my first ride compared to the others-- I think that's when the bike has just been pulled out of a warmer car.

You can avoid this problem by riding your bike for a bit just before calibrating-- or by carrying your bike outside the car not inside-- but this is just one example.

By the way I don't want credit for putting a course next to or very near a straight line; if I remember right this advice has been in the measurement handbook since way back!
I often find that my first cal ride is lower than the subsequent rides. But I carry my bike on the outside. Sometimes, I have to throw out my first cal ride because of the first ride is too far out of whack. But, experimenting with different ways of initializing the counter results in different readings for me. That is, the more zeros on the counter when I start, the more likely the first ride may be 2, 3 or more counts off of the other rides. I think this may have to do with the "gear slop" in the counter, but I also wonder if it has something to do with my 5-digit counter. Even when I do my utmost to be precise in every way, I consistently get this "first cal ride outlier phenomenon". Even when I roll forward as I get set for ride #1 - to the point where the gear in the counter is touching and the zero just begins to move, thus eliminating the "slop" factor, the counts tend to be lower than rides 2-4. I intend to get to the bottom of this eventually.
Lyman, maybe you can try not "setting" your counter before calibrating. Just roll forward to a whole number, write it down and start. I'll be curious to see if you still get that effect.

AJ Vanderwaal, who taught me about all of this, used to say that repeated calibration is one of the best ways to improve your steady riding for measuring. He also showed me how to count spokes for a more precise measure of consistency on those calibration rides.

To clarify, you don't really count them, you number each spoke. When you stop at the nail you read the number of the last spoke you've passed and then estimate in tenths how far to the next spoke. This is somewhat like Neville Wood's method for using a bike computer, only Neville divides the circumference into decimal parts rather than in 1/36ths.
My method was always:

Push forward until a nice start number is reached.

Lock the wheel with the brake and set it down on the mark. Record the count.

Ride to the next mark, lock the wheel and record the count. Maintain locked wheel.

Turn bike around with wheel locked and set it down on the mark.

Ride back to start and repeat until you have enough rides.

You will catch a few 1/2 counts, but should get pretty good consistency between rides.
My calibration course is about .42 miles from my house so I always ride to the calibration course on both pre and post. I very seldom vary by more than 1 or 2 counts over 1320 feet. I do every once in awhile will go three and will ride the cal course again if I am 3 counts. I have yet to figure out why that it but it is more than like sloppy riding on my part. I do it the way Pete says.
I have been doing it the way Pete describes, as taught to me by Bob Thurston. The differences are that I have not been in the habit of riding before calibrating, and I tend to pick up the front wheel and hand spin it to a round number. I am a 200-pounder, and my bike weighs almost 50 pounds. It would not surprise me at all that riding even a short while warms up the tire.

I am going to try riding for a few minutes, then roll to a round number instead of spinning the wheel. If this doesn't do it, I am getting a new counter.

Thanks for the good suggestions.
Lyman - If you get a lower count on the first ride it is unlikely that the counter is at fault. When a counter fails, the usual thing is for it to skip 10 or more counts. I can't see any way that the count could increase because of a counter malfunction.

Can you provide a four-ride example of what you consider to be unusual counting?
Thanks, Pete. This is still a little unresolved for me. The second course sent to me for certification since I became Maryland state certifier had the following 300-meter cal course rides:

Pre:
2629
2632
2628
2633

Post:
2600
2605
2598
2602

I told her these were a little too far out of whack. She pushed back, saying that it was windy, and the rides in one direction are consistent.
Remember, calibration is supposed to mimic the way you ride. If you get a bit sloppy on a calibration ride or two, it will have little effect on the overall length of the course.

Also, if you keep calibrating until you have four identical numbers, it is likely that they will be lower than the ones you discarded,and you will have effectively reduced the size of your SCPF because your layout constant will be lower.

Sending someone back to do the whole thing over is a pretty extreme action, and there is usually a better way to attack the problem - if the problem really exists, which it doesn't look like it did, at least to me.
Thanks for the comments, Pete, Mark, J.A.

I did go ahead and grant the certification prior to posting this, using pretty much the logic you cited - #1, it wouldn't affect the working constant, and #2, rides in the same direction were consistent.

The value of this for me is to get a strong grasp of when to be concerned and when not to be. Thanks.

Add Reply

Post
×
×
×
×
Link copied to your clipboard.
×