Way to reduce lateral movement of a rod when trying to calculate the mass moment of inertia using bifilar pendulum? Basically I'm trying to measure the moment of inertia of a hollow rod using bifilar pendulum. I will rotate the rod, and measure the period, and use the written formula to calculate the moment of inertia. 
The problem is that I cannot seem to remove the lateral movement of this rod when I push it to the side on one end. 

This is how my rod looks like. It is a hollow cylinder that will rotate about its y-axis. However, when I give it a push to the sideway to rotate, (that is, rotate around the y-axis) it will move back and forth in x-axis direction and I'm worried if this will affect my experimental result.
Is there any good and creative way to reduce/remove this annoying back and forth motion of the rod without inputting another possible variable to consider in the experiment?
Thanks for any help in advance.
(P.S. I'm new to this forum, so please understand if I did not follow any formats or whatever :) )
 A: 1st Suggestion
Drill a narrow vertical hole through the middle of the tube and insert a steel wire so that it is aligned with the y axis. Then the rod will rotate about this fixed axis and not move in the x direction.
However this might be more trouble than it is worth :


*

*it could be difficulty to drill, getting the hole exactly through the axis of the tube, and avoiding damage to the shape of the tube;

*the increased friction will reduce the number of oscillations which you can time and also reduce the frequency of oscillation; you might be able to estimate the damping factor from the exponential decrease in amplitude (assuming you are able to measure it - see Note below), then use the damping factor to make a correction to the frequency of oscillation;

*drilling a hole might spoil the use for which the cylinder was intended.
2nd Suggestion
Glue very light but strong threads at both ends of the tube on opposite sides. Trail the threads horizontally and perpendicular to the stationary tube in the equilibrium position. Then redirect them through smooth tubes to a common point where they are glued or tied. 
Set the measurement tube in the starting position, displaced the same amount each side. When ready burn through the thread near the midpoint to release the tube so that it starts swinging. 
Provided the thread is very light it will have little effect on the swinging of the tube even though it is still attached. 
Note : 
A useful way of measuring accurately both the amplitude and period of oscillations is to glue a very light mirror to the midpoint of the tube. Direct a laser at the mirror and mark out a scale where the reflected spot moves across the wall.  
