The mysteries of the force curve

For OTW rowing, I am not even convinced a force curve should be looked at independently from boat speed and acceleration. On the other hand, wouldn’t it be great if we could extract a few meaningful parameters from the force curve to help analyze technique?

The approach Nielsen-Kellerman have taken is using “Slip” and “Wash” parameters, which basically tell give you a reduction in effective stroke length. Slip is basically the stroke length (or oar angle difference) it takes to get to 100N of force from the catch, and Wash is the oar angle difference (or stroke length) near the finish at which the force drops below 100N. The figure shows this well (from the brand new “stroke profile” plot on

Schematic Stroke Profile from an NK Empower Oarlock

It turns out that the schematic stroke profile is pretty close to the real one. When I calculate the area under the curve and compare with the average stroke force, there is pretty good consistency.

The ratio of average/peak force is a value between 0.5 and 1, the latter value just being a theoretical upper bound for a perfect block shaped force profile and the former the value for a triangular force profile. My values are pretty consistently around 0.57, but Greg Smith’s values are significantly higher (above 0.65 if I remember well).

So in the name of science, I did the following experiment on the erg. I must say using Painsled for iOS made it very easy to capture stroke profiles. I rowed a just row 150m off, 10 strokes “on”, varying stroke rate between 20 and 36spm. During the “on” strokes I tried to row what I consider good technique. Immediately after the 10th stroke, I grabbed the phone, did a screen capture, and took a picture of the PM5. I went up to 36spm, and back down, to see if there was consistency.

Before I show you all the raw data, here are the aggregate plots made on

As you can see, my Average/Peak force ratio correlated positively with stroke rate, being 0.56 below 24spm, and 0.60 above 32spm. Drive length was pretty constant. I didn’t see a difference between going up the ladder or going down. Work per stroke correlated positively with the stroke rate, as did both average and peak force.

So here are the raw data, one set of stroke rates per picture gallery:

20 spm





The graphs from Painsled show the last stroke of the set of ten in bold, and a few of the strokes before in a lighter line. The pictures of the PM5 show the final stroke. I was struck by how different the pictures look from the graphs, but I guess that’s an effect of the vertical scaling.

I will have to think about this a bit more. Here is a link to the article by Kleshnev who has numbers between 0.5 and 0.6, with 85% of the rowers at the higher end.

I guess the next step will be to film my stroke.

At the least, I think this post is a good demonstration of the power of a good data analytics site, combined with the excellent erg data capture of Painsled.

Post Scriptum

Discussing the stroke profiles, Greg and I made the cumulative plots of the Average/Peak force ratio on Here are the two plots. Interesting to see the differences.

The short URL of the present article is: