(a) The diameter of a thread is so small that it cannot be measured using a meter scale. We wind a number of turns of the thread on the meter scale so that the turns are closely touching one another. Measure the length (Z) of the windings on the scale which contains n number of turns.
Diameter of thread =1/n
(b) Least count=$\frac{pitch}{number of divisions on circular scale}$

So, theoretically, the least count decreases on increasing the number of divisions on the circular scale. Hence, accuracy would increase. Practically, it may not be possible to take the reading precisely due to the low resolution of the human eye.
(c) A large number of observations (say, 100) will give more reliable results than a smaller number of observations (say, 5). This is because the larger the number of readings, the closer is the arithmetic mean to the true value and hence the smaller the random error.