A couple years ago, I expressed reluctance to measure a 30 km because 11 miles of it was on a dirt road. Then & now, I feel uneasy w/how accurate any of the miles would be on those dirt roads. Others on this board said: go ahead and measure it using our system because it's still the best way. That I knew but I left the measurement to somebody else because I lacked confidence in how long each mile would end up- feeling each would be short. How short would vary from the consistency of any particular stretch of the 11 miles of roads.

Now I have a measurer asking me if it's okay to measure a marathon on dirt roads. Based on the previously mentioned reaction of those on this board, I've told him it's okay to measure it. I explained my misgivings but it's fine to get on w/the job.

He also asked if he should lay out his cal course on dirt. I don't know the answer to that although I lean toward a paved course. I think that's what the other guy used above (that ended up measuring the 30 km). This way, he could come up w/reliable, consistent cal numbers. I fear a dirt cal course would yield inconsistent numbers and larger than on pavement. Plus, there's the problem of picking out a 'representative' stretch of dirt road for the cal course.

Should the guy use a paved or dirt cal course to measure the dirt road marathon?
Original Post

Scott,

I agree that a measurement on a dirt road will have less accuracy than we like. However, the course will normally be long, if you calibrate on a paved cal course. If you would normally have 18000 clicks per mile on a paved course, each mile will be long, due to slippage (moving forward).

It would be more accurate to have the cal course on a portion of the course that is to be measured. That way, you will compensate for the slippage on the dirt. If one would normally have 18000 clicks per mile on pavement, it may turn out to only have 17950 clicks per mile on a dirt cal course. (How cruel - our measured course may be a bit long on a dirt/gravel course, and runners lose efficiency on dirt, as they lose some length-of-stride due to shoe slippage on the push-off. They end up taking more steps ech mile, and the course is long.)

This presents a good opportunity for comparison. Have the measurer set up a cal course in the dirt. After his post-cal rides, if he can go to a paved cal course and do 4 rides, we all can see what the difference is between the two surfaces.
The huge variation in what someone might call a "dirt" road would make me relunctant to assume anything about whether a "dirt" cal course would result in bigger or smaller numbers than a paved cal course. For that reason, if it was me I'd try to calibrate on a surface that is as representative as possible. But that brings up another issue. How accurate is his dirt cal course going to be? It's much easier to accurately measure out a cal course on pavement than on dirt, especially if you are doing it by yourself.
I was assuming the cal course would have to be set up with two people, since putting tension on the tape would be next to impossible with only one person. I doubt you could properly drive an anchor into the road with any degree of accuracy. Of course, you could note the measurement point, drive the anchor short of the measurement point, then make a note of where the measurement point is on the tape, whether it is at 4.25", 6", etc. Could be done, but not efficiently.

I will question the degree of disparity you have between calibration and the rest of the course. Even if the dirt surface varies, how much difference will there be? A foot per mile? Two feet? I imagine it would still be within our SCPF, as some sections may have more, and some have less slippage.
I'd stay away from this.

I had a couple of experiences last year that have made me extremely gun-shy about measuring on anything but paved roads. One was on grass, and the other on a thawing crushed limestone path.

I think you can assume that the tire is going to deform the same way on hard surfaces like asphalt and concrete. But if you calibrate on a hard surface and measure on a soft surface you're going to get bad results, because the tire deforms differently. You may get consistent results, but they will still not be accurate.

My results last summer showed 11.08 counts per meter on asphalt, and 10.99 counts per meter on grass fifteen minutes later. So a 5000 meter course would be 55040 counts using the asphalt constant and 54950 counts using the grass constant. The course I was measuring (on grass) was 52581 counts, which was 4785 meters long using the grass constant and 4746 meters long using the asphalt constant. So I would agree that using a hard surface constant to measure on a soft surface will usually err to the side of making the course long. The steel tape pegged the course at 4788 meters.

Should you decide to set out a calibration course on a soft surface, get some utility marking flags and use them at each tape end. And depending on how wide open your course is, a couple of traffic cones might help you stay on a straight line.
All:

To what do you attribute the lower constant obtained when calibrating on a non-asphalt course? Is it really slippage between the wheel and the ground? Wouldn't that make the soft course constant larger? Jay's soft course constant was smaller than the asphalt constant. Jay's soft course constant returned a course length pretty close to the pegged and steel tape measured course. He doesn't say whether he applied the prevention factor.

A dirt road surface, like on a country road that cars drive on, is pretty hard (when dry) and should provide a constant the same as an asphalt constant, assuming the rider doesn't allow the un-driven front wheel to slip (this must be skidding sideways due to the uneven surface). A crushed stone path or grass surface is something different. The un-driven wheel could sink in and measure below the surface. I haven't figured out what the affect on the constant would be for this condition.

Our Application for Certification (question 27) asks for the type of surface but really provides no help for determining whether the constant is impacted by the surface.

Unless the unpaved surface is frozen or hard and smooth, I think Jay is right, we really have no way to accurately determine the length of the course.

Pete
Guido Bros
The smaller constant is from the wheel not turning as many revolutions over the same distance.

Imaging a cinder trail (loose cinders, not hardpacked) next to a concrete bike trail. Make two marks on the concrete, separated by 50 feet. Ride the bike tire along the concrete. You may get 9 complete revolutions. Now, ride the bike on the cinder trail. You may get 8.75 revolutions.

While the tire is being pushed through the cinders, there is less friction with the ground to make the tire revolve. You can exaggerate the effect by applying slight brake pressure. When doing that, you can see the tire sliding through the cinders, instead of smooth rolling like you get on the concrete.

Since the tire is sliding forward part of the way, there are fewer revolutions for a given distance. Make sense?

I think that if you calibrate on the loose surface, and the degree of looseness is consistent for the course, you will get an accurate course. There is a definite threshold, though. A sandy beach is too loose for any consistency. That is where the measurer has to make the determination as to when the surface is too inconsistent to allow for proper measurement. Experience is the best guide.
Sorry Duane, I disagree. It requires very little friction force to make the front wheel turn. The only thing resisting is the inertia of the wheel, and if you are going a constant speed there's not even that. The threshold friction force required to prevent slipping forward is very low so I don't think it ever happens. The rear wheel is different of course, as you can easily overcome the friction of a dirt road by pedaling hard. But for the front wheel, there is no other significant force acting on it besides the friction.

What CAN happen on a surface with very low friction is that the wheel will not change its perimeter. For the perimeter to become smaller, there has to be enough friction force to overcome the tire's resistance to shortening in the contact patch. Mike Sandford believes that there is always enough friction to cause a "stick" condition in the contact patch and shorten the wheel perimeter, and I don't necessarily disagree with him. If that's true, then the cal constant will not change with different road surface friction coefficients.
Mark,

While it does require little friction to make the front tire spin, if you are riding on pebbles, sand, or cinders, the loose particles beneath the tire will slide forward and sideways, lessening the spin of the tire. I'm not talking about a hard-packed dirt surface, but a surface where there may be 1/8" to 1/4" of loose particles.

I will try to get out to a location that has both pavement and cinders next to each other, and do the demo I discussed in my earlier posting. I will posit that I will have fewer clicks on the cinders than on the pavement.

We shall see. Does what I am saying make sense? If not, why can't we certify a straight stretch on a beach? Arrow-straight one-mile, for instance? Calibrate on pavement, then measure on the beach, I am sure you would get a significant variance.
The issue as I see it is one of consistency. when you measure over a hard surface, it is uniformly hard. On the example I cited, though, there were a lot of different conditions present. Some of the course was grass. Some of it was dirt. Other parts of the course were either or- but extremely soft and hard to ride over.

I don't know whether or not my calibration course reflected the surface I measured.

The numbers that I quoted DO NOT include SCPF as we were trying to determine the length of an exiting course- much like a validation.

I really don't think that friction plays a big part in this. The front wheel steers the bike and holds up its front end. And if we want to certify a straight stretch on a beach, let's pull out a steel tape. It would be one hell of a lot easier than trying to ride a bike over it.

Based on my experiences over the last year or so, it is my opinion that if the course includes large segments where the tire leaves an impression, the right answer for those segments is to measure them with a steel tape.
Duane, when you ride through a puddle the water gets pushed in front of and to the sides of your tire. That doesn't mean the tire is slipping. No difference with cinders or pebbles or whatever.
In the May 1998 issue of Measurement News Mike Sandford published the first of four articles on calibration variation with surface roughness. The first was a survey of past literature on the subject.

It’s 388 kb in pdf format. Write to "riegelpete at aol dot com" and I’ll send you a copy. I have not yet found an online spot to park articles such as this. Tips welcome.

My view on this is that it’s desirable to calibrate on the same surface, but that if you calibrate on pavement, any non-pavement courses you measure will be a bit off-distance. This could cause a problem if the course is validated. However, I believe that requiring steel-taping will be counterproductive, as this is time consuming and courses will remain unmeasured.

I believe the great mass of runners are better served by courses that are a small amount off, as this will give them better courses than they will get if the course is not measured at all.

So, I’d say go ahead and calibrate on pavement and don’t worry about it. The non-pavement course will be a bit off, but not enough to make a serious difference.

Heresy. I know.
Duane:

I'm not sure I can visualize the slippage you cite. If the front tire axle is directly over the contact spot, what force causes the slippage? If slippage is forward motion without wheel rotation, the only way I can see that happening is through some force acting on the wheel in the opposite direction of travel. Friction at the contact spot does this. I'll have to look at Mike Sanford's articles if I want to understand the science here.

In the mean time, I think Pete Riegel has hit the nail succinctly on the head. I was helping a new measurer who wants to measure a marathon on a crushed stone walking trail. We set up a calibration course on pavement and measured a segment on frozen crushed stone. Then the snows came, it will be interesting to see whether he continues the measurement. However, I believe that measuring the course using a bicycle calibrated on pavement will result in the best accuracy obtainable. Wheteher ceritfication of this course should include an asterisk is another question.

Pete
Guido Bros
If I'm distilling replies correctly, calibrating on a paved/solid surface will yield a long measurement of a dirt road course. How long is long is unknown.

The scenario; measurer: "Yes, I can measure your course on dirt roads. It's going to end up a little long."

Affected race official: "I see. How much is a little long?"

Measurer: "Unknown."

Race official: "Really, how's that?"

Measurer: "(This is were I need help from the forum)."

What isn't clear to me is why we don't use a dirt cal course to measure the dirt road course. Or, to be inclusive, measure ANY portion that includes a dirt stretch that exceeds, say, half a mile.
Okay, Mark, them's fightin' words! (Your most-recent post.)

I had to get on my bike and test, to see who is right. Happily for my ego, my test results show I am correct.

I went to a stretch of trail where there is a concrete trail for bike and 'bladers, and an adjacent cinder trail for runners and walkers. The cinders are pretty-well packed, but still have about 1/8" of looseness on top. Perfect for this test.

I marked Start and End points on the concrete. Rode 4 times between the marks. Paved counts: 91.75, 91.76, 91.76, and 91.76 revolutions (I use electronic measurement, and have my rim marked). Then, rode the cinders, within 1' of the edge of the concrete. 91.61, 91.61, 91.62, 91.58 revolutions.

Av. rev. on pavement: 91.7575
Av. rev. on cinders: 91.605

Based on my most-recent calibrations on a paved cal course, 1200', with average 177.30 clicks;

Test course was 621.03 feet long

Clicks per mile on pavement: 780.12
Clicks per mile on cinders: 778.83

Yielding a marathon course on cinders that would be 229 feet too long. (No SCPF used for any of these calculations)

Which seems to prove, based on this test, that a bike calibrated on pavement, but with the entire course on cinders (or similar loose surface) would yield a course that is too long.

Another way to explain how the tire slides forward in loose material may be:

1) There is a force pushing the tire forward. When on a solid, high-friction surface, the tire rolls predictably. Each point on the tire that is "X" distance from the current point of contact with the ground, hits the ground "X" distance from that current point of contact.

2) When on loose material (cinders, in this case), while there is a point of current contact, as the tire is pushed forward, the tire also is moving forward while being carried by the rolling pieces of material.

Envision 1/16" ball bearings, 3 layers thick on a floor. If you push a tire with a 7' circumference forward through the ball bearings, you will not contact the ground exactly 7' forward with each complete revolution. The sliding of the ball bearings allow the tire to "slide" forward a little with each ball bearing.

Does this explanation make sense? I have shown that it is factual, but I am not a physicist, so I can't put it in long-hair terms. I just try to explain it in simple terms, and have the data to back it up.
Duane:

I think your data show that your wheel is smaller on the hard surface. The tire deforms more on the hard surface. On the soft surface, the ground "absorbs" some of the deformation.

I'm not sure about your ball bearing metaphor, once rolling, something has to make it stop rotating as it crosses the ground.

As for Scott's questions [How much is a little long and why don't you know how much?] , it seems that one answer could be:

"Maybe as much as 1% (1300 ft in a marathon) if I calibrate on asphalt and measure on dirt. We don't know how much too long because the accuracy of our measurement depends on our method of measurement. That method is based on calibrating on accurately measured calibration courses laid out on hard, flat, level roads that have consistent surface and texture over their length and are similar in texture and consistency to courses normally measured on roads."

Additionally, this discussion seems to indicate that calibrating on a surface similar to the dirt road race course would be a good idea. That assumes that the calibration course meets all the reasonable requirements; flat, level, free of holes, bumps, loose gravel and that the calibrator doesn't ride over large loose stones. All that said, it also seems that measuring on a "dirt road" introduces the un-repeatability of the shortest possible route. That seems to be a larger contributor to inaccuracy than the calibration.

Pete
Guido Bros
Bros,;

I guess we won't resolve the "why", just that measurements are different if calibrated on pavement vs. gravel, and measured on the opposite surface.

As for the ball bearings, my point is that the tire is not always in contact with the ground. The tire is supported by the ball bearings (or cinders) for a period of time, and distance, before the bearing/cinder is displaced and the tire contacts the ground. It is during that time that the tire doesn't continue its uniform rolling motion. It is supported, and additional cinders in front of the current cinder can inhibit the rolling of the tire, thus allowing the tire to move forward without an equal rotation of the tire, as the tire slides along on top of the cinders. Just like the Egyptians rolling their blocks of stone forward using logs underneath.
I think you need to determine what percentage of the course is on dirt or soft surface. I directed a 25k race today that has slightly less than a mile of hard-packed dirt; I don't even worry about the difference (the course is so damn hilly no one will ever set any record, personal or otherwise, on it).

I measured a marathon in So. Dakota that had a longer stretch of crushed rock on the Mickelson Trail. I could definitely feel my front wheel slipping a bit, especially on the steeper sections when I'd stand up to pedal and unweight the front axle a bit.
I agree with the above statement that says:
quote:
I'm not sure I can visualize the slippage you cite. If the front tire axle is directly over the contact spot, what force causes the slippage?

The problem is more likely to be apparent diameter of the wheel. This will be more true when the tire is soft. On a hard road you get deflection towards the hub as the wheel is supported by a small patch of road. On a soft surface more of the tire is in contact with the ground so the psi per sq inch is less, and therefore the tire may deflect less and the operational diameter of the wheel would be different, I think softer tires would probably be worse.

Of road courses often have a diffrent surface temperature than black top courses. This temperature gets to affect the whole wheel. Temperature affects both air presure and the wheel diameter.

When repeating doing very accurate and repeated runs on a road bike, with tire pressure near 100 psi, a simple thing like a rain shower that leaves a few puddles on the road will change my counts on the next pass. I would guess the surface temperature of the off road course would similarly alter the counts per mile.

Back to slippage, unless the grass was so wet you were sliding around, or your front hub baring was rusty and sticky, I think slippage would be imaginary.
It is amazing how much interest this topic has raised. I think the bottom line is how confident the measurer is in the distance he has measured.

Last year I measured a marathon and a half marathon on Vancouver Island. The marathon was fine, but the half had 4km on a trail. When measuring that stretch my front wheel slipped noticably. In the end I told the organizer that I was reasonably confident the half marathon distance was accurate, but I was not sufficiently confident that I would apply for certification.

On the other had I measured a half marathon a few years ago that was mosty on a dike with hard packed gravel. I was confident that in my measurement and very confident that if there was an variation it was less than I would see with a temperature change. For that event I laid out a calibration course on the dike and compared it with a calibration on pavement. The variation was small. This was my file note for that measurement:

A calibration course was established on the gravel path and the bike constant measured on a certified calibration course (BC-2002-004-BDC) for comparison. The calibration constant on the bike path was 12,000.3 compared to 12,004.5 on the Colony Farm calibration course. The difference is small and since 90% of the route is on the gravel bike path that constant was used for the entire course. Given the difference in calibration constant and the small amount of paved path the error would only be about 0.7 m in the half marathon.

Paul
In general, a rough surface will get from 1 to 5 SCPF's more than a normal asphalt surface. In other words, if you lay out a dirt road 10k using your regular asphalt calibration course, and do a perfect job of riding, the resulting course will be around 10 to 50 meters oversize.

Nobody has come close to pinning it down any better than that. The problem is that dirt roads have a surface roughness that varies much more than do the various asphalt and concrete surfaces. How do you pick a stretch to calibrate on and be sure it fairly represents the rest of the course? I don't know.

I suspect that a dirt-road cal course would be better than an asphalt one, but I would not force a measurer to do it.

When I was running in the 70's, I was never so consistent that my times could be used as a ruler, as I have seen several top runners claim they can do it. When I've read that Joe Hotshoe could tell that the course was 30 or 50 meters off my reaction is "bull." When certified courses came in I was glad to have something better to use. I believe a dirt road measured with an asphalt calibration course is so much better than what went before that I don't worry about whether it is spot on. It's plenty good enough for a guy like I was to measure his effort. All courses of a nominal length vary in length from one to another.

I would hate to see us get tied up in picky rules and regulations. If we make it too much of a pain to measure, it won't get done. That would be more of a bad thing than a few courses that are mildly oversize.