# Possible cause for Google Maps / bicycle odometer divergence (tire size setting correct)

I have plot my exact route from home to my office using Google Maps. By exact, I mean that I took nearly every turn and change from auto-calculated route to match my exact route into consideration.

Google Maps claims that my route is 10.4 km. Everyday I'm riding it, my bicycle's odometer claims that I have actually traveled 11.11 km.

I did exactly the same with my return route (a bit different) and after including every actual turn, change etc. it turned out that route is 9.5 km according to Google Maps and 10.43 km according to my odometer.

The difference is 0.7 / 11.1 = 6.3% on the outward trip and 0.9 / 10.4 = 8.6% on the return trip1.

My first and of course obvious clue was an incorrect tire size setting in my odometer. However, I double-checked to be sure it's correct:

I have (50-559) 26x1.95 tires, so I believe I should set it to 2089.

What else could cause the difference?

I have consulted this with a very experienced bicycle rider, but he mistakenly thinks it's because the Google Maps route doesn't always match the actual route. However, I spend some time adjusting the auto-calculated route, so it matches my actual route in about 99.5%.

1 Assuming that my difference calculation formula is correct. Plus -- I don't know why divergence is higher in case of return trip. The only clue, that I have is that it takes much more backroads, turns, slow downs (while onward trip is riding on the streets nearly all of time), which causes, that travel time is longer in case of return trip even though it is shorter. But, that is a weak argument as divergence should be the same in both / all cases.

• Try method B to determine the wheel's circumference. You also have to consider that the dynamic diameter of the wheel is different from the static diameter caused by tyre compression under load. – Carel Jun 18 '16 at 8:13
• So when you ride you go perfectly straight and don't weave back and forth at all? And have you done a roll out on your tire to measure it's actual circumference? (And, actually, 6.8% accuracy is pretty darn good.) – Daniel R Hicks Jun 18 '16 at 11:21
• Maybe you could post/share the Google map route which you're asking about? – ChrisW Jun 19 '16 at 0:01
• @ChrisW It's in Poland, so it won't tell you much. Plus, as said above (Daniel R Hicks' comment) ~7% isn't that bad at all. – trejder Jun 19 '16 at 18:59

I point out that your comment of

`````` I have (50-559) 26x1.95 tires I should set it to 2089.
``````

Doesn't actually match your table, which says:

`````` 50-559 (26"x1.9")     2089 mm
54-559 (26"x2.0")     2114 mm
``````

So the ETRTO number and the imperial measurement are not quite equal. Could be your circumference is closer to halfway between these, or 2101 mm

Downside, this would put your reading up by 0.5% making it even larger, and further from the map distance. Do the rolling distance measure repeatedly as others have suggested, for an accurate figure.

You can also consider putting your reader-magnet on the rear wheel, which should have less slip than the front one, and will follow a shorter path while you steer.

• I do wonder if the OP rides fairly slowly and weaves a lot with the front wheel, whether that might add 5%. It seems unlikely, but you never know. – Móż Jun 18 '16 at 11:48
• But if the OP pedals hard there may be tyre slip on the back wheel. – Nuі Jun 19 '16 at 2:35
• @nui Sorry I wasn't clear. By tyre slip I meant what Móż saw, the back wheel will always track a straighter line than the front wheel and will therefore take a shorter path. You're thinking of the rubber slipping on the road, making it look like the wheel has gone more than a full revolution. If that amount is measurable with common tools I'd be surprised! That would be the same as sidewall deformation seen in a dragstrip car, and would be so small as to be approaching 0 on a bike. – Criggie Jun 19 '16 at 3:49
• @Criggie Great answer, thanks. BTW: The idea with putting reader-magnet to the back wheels sound great. – trejder Jun 19 '16 at 19:04
• @Móż This assumption could work for my back route. As for riding to work, I'm travelling 99% on streets, having amount of weaves reduced to minimum and having speeds on approx. 20 km/h average and up to 35 km/h maximum. So, I assume, this doesn't count as "fairly slowly". – trejder Jun 19 '16 at 19:06

Google maps probably treats roads as not having any width, and corners as point turns. Unlike bends, if you zoom in you tend to see a sharp 90 degree change in line on a curve This will cost you the length of the arc you really take.

It certainly omits all the lane changes, obstacle avoidance and similar manoeuvres.

However there is some variation in tyres - particularly some of the anti-puncture tyres have a bit of extra thickness; inflation vs. weight on the wheel will also have an effect. So you're better off seeing the wheel circumference based on counting the wheel revolutions over an accurately-measured straight line.

• I don't think you have thought this through. Lane changes could add 1% to the trip, but if Google is measuring corners as right angles, it's measurement would be longer, not shorter.The two effects would tend to cancel. (Circumference of a circle is 2PiR, about 6.3R. For a right angle corner, the quarter circle path is 1.57R while the "square" approximation is 2R) – andy256 Jun 18 '16 at 8:56
• @Andy256 I implicitly assumed a road of negligible width on the map, so 2R_map <<2piR_real. Effectively this would be like the difference between following the kerb on an inside corner (left in the UK, right in the US etc) and following the middle of the lane. Of course to do the corner you'd have to stop, pick the bike up and turn it 90 degrees, then start again. – Chris H Jun 18 '16 at 9:12
• Maybe I didn't explain myself clearly. This "square corners" approach should mean Google gets a longer distance. The OP is seeing the opposite. Hence this is not the cause. – andy256 Jun 18 '16 at 9:29
• @Andy256 I get that but I'm saying R!=R for the two cases in your first comment. The location of the narrow road (or narrow carriageway) will also have an effect. – Chris H Jun 18 '16 at 9:57
• I think i see the difference. I'm isolating a corner, you're integrating over the whole route. Your approach should be more valid though I'm not sure with a system that jumps to roads. – Chris H Jun 18 '16 at 10:03

First do a rolling measure while sitting on the bike to get the speedo calibrated to your actual setup. That's your method B

Then I suggest also getting a proper map and making sure google's scale is correct. They care about navigation much more than 5% errors on inner-city distances, so it might be that their map is less accurate than a proper cartographical one is. Elevation change could likewise make a small difference.

The small weaving motions you make while riding to stay balanced also add to the distance, especially for the front wheel. That's likely to be worse the slower you go and the more you weave around other road users. In the worst case, if you're weaving through stationary motorists at rush hour you could easily add 5% just for that. Similarly if you're travelling barely faster than walking on a crowded shared path.