@ Tanuja , it is quite straight.
First step. take mean of x variables.
compute (xi-x mean) for each element.
Same things for y variable also.
Take mean of y variable and compute (yi-ymean) for each element.
Now, multiply the two values for each corresponding data point.
That mean if you take first data point for x ( xi-x mean) where x=1
then it should be multiplied by (yi-y mean) where I=1 for y value.
Then take the sum of this each data points product ( xi-x mean) *( yi-y mean).
Finally, for population covariance just divide the sum by N( n is the total observations).
For , sample covariance just divide the sum by (n-1)[n is the total observations).