(a) The thread is far thinner than the smallest division (1 mm) of a metre scale, so it cannot be measured directly. Wind the thread closely on a uniform pencil so that the turns touch one another, and count the number of turns n. Measure the length l of the wound part with the metre scale.

Each turn occupies a length equal to the diameter of the thread, so
diameter of thread=nl
Winding many turns makes l large enough to measure with the metre scale.
(b) Least count =number of circular divisionspitch=2001.0 mm=0.005 mm.
In principle, more divisions give a smaller least count. In practice the accuracy cannot be increased arbitrarily: the divisions become too close for the eye to tell apart, and it is impossible to make the pitch of the screw perfectly uniform along its whole length. So increasing the divisions beyond a limit does not improve the accuracy.
(c) Each reading contains a small random error, which is equally likely to be positive or negative. When many readings are averaged, the positive and negative errors largely cancel. With 100 readings the cancellation is much better than with 5, so the mean of 100 readings is a more reliable estimate of the true diameter.