I conducted a simple experiment where I tried to find the relationship between a cylinder's mass and its time taken (or average speed or acceleration, which can be easily derived from the time & distance).

I rolled a cylinder down a fixed ramp (length & angle-wise), varied the mass of the cylinder by filling the container with various objects that provided even mass distribution and measured the time taken for it to reach from the top to the bottom of the ramp.

I have obtained a result of 10 data points from 10 repeats each (values ranging from mass 0.073kg & time 2.119 to 2.278kg & 1.835), which, when plotted (time in y-axis), shows me a clear trend that looks like either a declining natural exponential or the positive side of an inverse square curve. The following two functions fit the data values most accurately:

t =0.7298 * e^(-13.14*m) + 1.842 t = 0.001644 * m^(-2) + 1.850

time-mass graph

Here, I couldn't figure out what the relationship between mass and time (or speed; acceleration) is, and what the supporting theory and the relevant equations are.

Now, the law of conservation of energy does not explain such a relationship, as mass does not affect the velocity of the cylinder, i.e. GPE = ∆KE(translational) + ∆KE(rotational) mgh = 0.5mv^2 + 0.25mv^2 , where m is cancelled out.

Another possible contender is the rolling friction of the cylinder, but from my knowledge and from what I have tried, this also cannot explain the relationship between the mass and time.

The most plausible explanation, I believe, is to do with air resistance; an object of greater mass experiences greater gravitational force, hence it takes longer for the air resistance to balance out this force, and so it accelerates for longer until it reaches its terminal velocity (much like why an elephant would fall much faster than a feather when there is air resistance). My current knowledge tells me that air resistance is proportional to the object's velocity. However, my research so far has told me that there is not really a definitive equation that links the two variables or explain the relationship between my two variables, as there are multiple factors that contribute to air resistance, e.g. surface area. Some sources suggest that it is:

F = Kv^2 or F = Kv, of air resistance F, velocity v, and an unknown constant K.

Although I do feel like air resistance is the reason behind this time-mass relationship, I can't really seem to find a link between this and my graph.

My supervisor has advised me to simply explain, in my investigation analysis write-up, the nature of the relationship and how it changes over time, possibly describing separately the two sections of the graph that differ distinctly in their gradients due to the rapid change at a point (due to its exponential or inverse square nature) - without using an equation to derive an explanation, if I can't find one.

However, I feel like the relationship between the two variables follows a pattern (whether it be exponential or inverse square) just too accurately to be unexplainable or accidental.

Apologies if I have rambled on for too long. Here's a summary of my problem:

  1. What theory explains the relationship between the mass of a cylinder and the time taken to roll down an incline?
  2. What equation(s) relate the two variables and explain the shape of the graph I have obtained? (either one the two functions stated above)
  3. Is it air resistance the reason behind the relationship, as I have mentioned? And if it is, is there actually an equation to explain the shape of the graph? And if there isn't, what would be the most effective/correct way to interpret and analyse such a relationship without using an equation?
  4. Or is the relationship I have really inexplicable through a single equation? Or is it possibly accidental, or a result of an experimental error that I do not recognise?

Please let me know if you have any questions about this or want any additional information that I may have omitted - which is highly possible, as this is my very first time asking a question on physics stack exchange.

  • 2
    $\begingroup$ How thick is the surface of your cylinder-container and what material is it made of? There could be discrepancies due to differences in moment of inertia $\endgroup$ Commented May 19, 2017 at 16:58
  • $\begingroup$ Exponentials are characteristic of $F=kv$ type force laws. If k is independent of mass, then the exponent has a mass parameter. You have discovered two things under the assumption of rolling: a] The force law due to air resistance is like $F=kv$, b] The constant k is independent of mass(or depends weakly on mass), since you have a strong mass in your exponential. I am sure you can work out the details of the time traversed using standard mechanics assuming a $F=kv$ force law and see if that fits your data. Thats a fun exercise in mechanics. $\endgroup$
    – Anonjohn
    Commented May 19, 2017 at 17:39
  • $\begingroup$ Another suggestion/extension of your experiment, if this fits the $F=kv$ data is to check what $k$ does depend on. Does it for example depend on surface area? you could change the length of the cylinder. Does it maybe depend on the specifics of the problem? maybe the acceleration/ other forces etc. You could just drop your cylinders and see what happens, without all the inclined planes. If you get a consistent value of $k$ in all such experiments, it would be time to celebrate! Physics works. If you don't, the Physics SE is a nice place to take the discussion forward. $\endgroup$
    – Anonjohn
    Commented May 19, 2017 at 17:45
  • $\begingroup$ Please provide details of the cylinder and its contents : inner and outer dimensions, mass of cylinder without contents, masses of contents. I agree with Abhijeet that the difference in mass distribution is a more likely explanation than air resistance. $\endgroup$ Commented May 19, 2017 at 22:39
  • $\begingroup$ @AbhijeetMelkani I believe the material of the cylinder is some sort of compressed cardboard, which I recycled from a teabag container. The thickness would be approximately 1-2mm. Radius is about 5cm and its height around 16cm. The mass of the empty cylinder is around 0.07kg, and I have used various materials such as cotton wool, iron wool, small grains, and play-doh, which vary in mass, which can be seen in the image I have attached in my question if you subtract the mass of the empty cylinder from it. I hope this information helps. $\endgroup$ Commented May 20, 2017 at 2:23

2 Answers 2


The most likely explanation for the variation in time of descent is the changing moment of inertia (MI) of the cylinder. MI changes because the distribution of mass about the axis of the cylinder changes.

Air resistance is frequently offered as an explanation for discrepancies in school mechanics experiments, but this is rarely appropriate. See my answer to Air resistance for practical purposes?

The acceleration down the incline is [1]
$$a=\frac{g\sin\theta}{1+k}$$ where the MI about the centre is $I=kMR^2$. Acceleration is constant so the distance and time along the incline are related by $s=\frac12 at^2$. Therefore, for a fixed length and angle of incline $t^2$ should be proportional to $1+k$.

The mass of the empty cylinder is distributed mostly at the rim. For such a hollow cylinder the MI is $MR^2$ so $k=1$. For a solid cylinder of uniform density the MI is $\frac12 MR^2$ so $k=\frac12$. The ratio of times of descent should be $\sqrt{\frac{1+1}{1+\frac12}}=\sqrt{\frac{4}{3}}=1.1547$.

Using your figures, and assuming the lightest cylinder approximates a cylindrical shell while the heaviest approximates a solid cylinder of the same radius, this ratio is $\frac{2.119}{1.835}=1.15468$. This is embarrassingly close to the prediction, and (I think) confirms that this explanation is most likely to be the correct one. There is no need to invoke either air resistance or rolling resistance as an explanation.

For a more thorough analysis of your data, $k$ has to be related to the distribution of mass in the cylinder.

Assuming that the empty cylinder has no ends, its MI is [2] $\frac12 m (R^2+r^2)$ where $R, r$ are outer and inner radii. The MI of the filling is $\frac12 (M-m) r^2$ where $M, m$ are total mass and mass of the empty cylinder. So the total MI of the filled cylinder is
$kMR^2=\frac12 m (R^2+r^2)+\frac12 (M-m) r^2$
$k=\frac12 \mu (1+\rho^2)+\frac12 (1-\mu) \rho^2=\frac12(\mu+\rho^2)$
where $\mu=m/M$ and $\rho=r/R$.

You should find that $t^2=\frac{L}{g\sin\theta}(2+\rho^2+\mu)$ where $L$ is distance of travel along the incline.

Using your data I plotted the following graph of $t^2$ vs $\mu=m/M$, along with a trendline, but you could use the values in your experiment $(L, R, r, \theta)$ to plot the predicted variation. Note that the data points are in reverse order compared with your list.
enter image description here
The data point for a mass of 0.196 kg $(\mu \approx 0.37)$ is an obvious candidate for investigation. My guess is that this might be the grains. Like a fluid, these would tend to keep their position while the cylinder shell rotates around them. Effectively the grains inside the shell are sliding rather than rolling down the incline; only the shell is rotating. This reduces the time of descent.

The measured time of descent is higher than my prediction when the mass of the filling is small $(\mu \ll 1)$. Possibly this could be explained if the container has cardboard ends.

[1] http://www.phys.ufl.edu/courses/phy2053/spring12/lectures/PHY2053_03-15-12.pdf, slide 6.
[2] http://hyperphysics.phy-astr.gsu.edu/hbase/ihoop.html


I went ahead and found the equation of motion of a cylinder rolling down a ramp, neglecting rolling friction$^1$ but including quadratic drag. Quadratic drag means that the drag force is proportional to $v^2$. The equation of motion is as follows, where $x$ represents the total distance traversed down the ramp:

$$ \ddot{x}=\frac{2}{3}g\sin{\theta}-\frac{\rho A C_d}{m}\dot{x} $$

I found the first term by Lagrangian mechanics, setting $$ T=\frac{1}{2}mv^2+\frac{1}{2}I\omega ^2 $$ $$V=-mgx\sin{\theta} $$

For a cylinder, $I=\frac{1}{2}mr^2$ and $\omega = \frac{v}{r}$.

Solving the Lagrangian equation

$$ \frac{d}{dt} ( \frac{\partial L}{\partial \dot{x}})= \frac{\partial L}{\partial x} $$


$$ \ddot{x}=\frac{2}{3}g\sin{\theta}. $$

Next, I just included the quadratic drag term $F_d=\frac{1}{2}\rho v^2 C_d A$, and divided by mass via $F=ma$.

Since $v^2=\dot{x}^2$, we can substitute that into the equations. The rest of the terms do not depend on mass ($A$ is related to mass via $\rho_{cylinder}$, but the $\rho$ in the equation is $\rho_{air}$)

Solving this differential equation, setting $a=\frac{2}{3}g\sin{\theta}$ and $b=\rho A C_d$, we get

$$ x=\frac{a}{b}mt+\frac{c_1 m}{b}e^{\frac{-bt}{m}}+c_2. $$

Using this equation, and setting $x(0)=\dot{x}(0)=0$ (since $x$ is distance rolled down the ramp), we can obtain $$ c_1=\frac{a}{b}m $$ $$ c_2=-\frac{am^2}{b^2} $$

which gives us a final equation of

$$ x=\frac{a}{b}mt+\frac{a m^2}{b^2}e^{\frac{-bt}{m}}-\frac{am^2}{b^2}. $$

This is very difficult if not impossible to empirically solve $t(m)^2$, so instead I just graphed it in desmos for arbitrary values of $x$, $a$ and $b$, just to see the dependence of $t$ on $m$.

This is the dependence I found, which appears to match your dependence very well! Included are residuals, though I did not include the error bars.

enter image description here

enter image description here

And here is the link to play with it yourself: https://www.desmos.com/calculator/akux3vsubk

I hope this helps / answers your question!

$^1$ If rolling friction is included, it just gets absorbed into the $a$ term, as it depends on mass, and so the mass cancels when finding acceleration.

$^2$ Setting a=b=x=1, we can solve for $t(m)$ and obtain (via Wolfram Alpha)

$$ t=\frac{m^2 W(-e^{-1-\frac{1}{m^2}})+m^2+1}{m} $$

where $W(x)$ is the product log function, or Lambert W function, giving the inverse of $f(z)=ze^z$

  • $\begingroup$ The maths in your answer seems to approach the question very logically and successfully, but I'm very sorry to tell you that I don't follow some part of the progression of your explanation. I currently am studying at an IB HL level, and so I may be short on some theories and equations you have used. I am kind of lost at the part where you solved the Lagrangian equation, which seems to involve some sort of calculus. Would you kindly explain briefly each step you took to get to the final formula please? $\endgroup$ Commented May 20, 2017 at 3:47
  • $\begingroup$ Sure! Yeah, I skipped some steps. The Lagrangian is defined as L = T - V, where T is kinetic energy and V is potential energy. So I just took T and V as stated and found L, and then took the partial derivative of L with respect to x and x' (and then the total time derivative of dL/dx') and used the equation (d/dt)(dL/dx')=(dL/dx) to find a final equation of motion, which gave me the x'' = (2/3)g sin (theta) equation. Here is a good pdf on the subject: www3.ul.ie/wlee/ms4414_lagrangian_mechanics.pdf $\endgroup$
    – Pawr
    Commented May 20, 2017 at 19:03
  • $\begingroup$ I now fully comprehend the formulation of the equation of the cylinder's motion. I am not quite sure of the quadratic drag form Fd, but I assume this doesn't have much significance in your explanation as a whole. Although, could you possibly reference the source of the formula and tell me what 'Cd' is? Also, more importantly, I don't understand the steps you took after this that leads to your conclusion. Could you explain how and why you solved 'the' differential equation (what is it, by the way?) incorporating substitution of a and b? I don't follow where exponential equation came from. $\endgroup$ Commented May 21, 2017 at 8:34
  • 1
    $\begingroup$ How well does your graph fit the experimental data? $\endgroup$ Commented Nov 14, 2017 at 2:37
  • $\begingroup$ $R^2$ of 0.9143 by fitting the free parameters. I'll edit in the graph of the data with the best-fit line plus the residuals. Seems pretty good $\endgroup$
    – Pawr
    Commented Jul 3, 2018 at 20:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.