You are on page 1of 2

The value of g thus obtained from the two inclined slopes, one of 15 cm and the other

of 30 cm rolling distances, are ________ and _______ respectively.

As explained above, the model applied in this experiment follows the following equation
for a rigid cylinder:

g
l= sin θ t 2
3

Since, in addition, the values that have been changed in each test have been the
angles, it is possible to rearrange this expression to obtain the following linear equation
to be used as a model to fit the data:

y=m x

Where

2 3l 1
y=t , m= , x=
g sin θ

In this way, for each angle, a different time has been obtained. Therefore, it has been
possible to get the g value from the slope by taking into account its error too. Hence,

g=
3l
m
, Δ g= | |
∂ g 3l
= Δm
∂ m m2

In order to fit the data, chi-squared analysis has been used, and the results are shown
below:

Rolling distance Chi-squared Red chi-squared P-value


15 cm 9.538 1.060 0.3891
30 cm 24.06 2.673 0.004203

−2
g15 =9.5± 0.3 m s
−2
g30 =9.5± 0.5 m s

As we can see, both from these values and from [graph 1], the value obtained for
gravity is the same in both cases, except for the fact that the error is larger in the
second case. However, the actual value of g is within this error range, so we can say
that the experiment is reliable although not completely accurate since the error is only
one order of magnitude lower than the value of gravity. There are several factors that
could influence this fact.

According to the chi-squared criteria for a good fit [table 1], that is, reduced chi-squared
around 1 and p-value around 0.5, the model just holds entirely for the first method,
whereas in the second, both values aren’t within the ranges of what can be considered
a good fit, despite the value for gravity is a good one. This means that is not a good
model for the second case, so let us look closely at it.

From the residual plots [graph 2], we see that both are very similar, not having any
visible trends. The points are randomly distributed around the horizontal axis, so a
linear regression model is appropriate for the data in both cases. Therefore, the linear
model we should apply to the second case would have to be a different one from the
first one. At this point, what could be the reason for that?

So far, it has been assumed that the cylinder does not slide in order to apply eq [3]. So
what could be the differences between the two planks? The first one (l = 15 cm) was a
bit rougher than the second (l = 30 cm), since the latter was polished. In principle,
according to the theory, the amount of friction does not matter, since the friction is the
condition so that the cylinder could roll. It is possible that due to the smoothness of the
polished plank, the cylinder slipped a bit on that one. We even recorded a slo-mo video
to see if it slipped, but it was not conclusive. We might have used an even larger
amount of frames per second, which was not at our disposal.

This fact could also be due to bigger air friction since the cylinder travelled further, and,
as it was made of wood, its density was not as big as if it were made of metal. It is
possible that random and systematic errors also took part in this process, but as these
concern the act of lifting the ruler (which could give the cylinder a little push), the
voltage readings from the circuit (which each reading is taken 208 microseconds after
the other, a value that has been neglected), maybe a little deviation of the cylinder’s
trajectory (which is longer in the second case), they apply to both cases. Therefore,
they are the origin of the 0.3 and 0.5 uncertainties.

A way to improve the experiment would be to use a metal cylinder and the same
wooden material in both cases, to limit the uncertainty about systematic errors.
Perhaps an automatic trigger to lift up the ruler and break the first connection would be
needed to avoid those initial pushes.

You might also like