How do I calculate the error in gradient????

put your values in a table mode and than your will get the equation of the line. Than what you do is get the correlation(R) of the line from your calculator. R is between 0 to 1. 0 to 4 means weak relationship. 0.5 to 0.6 means moderate and 0.7 to 1 means strong. If R is -ve it's negative weak/strong +ve means positive weak/strong.

Therefore, the closer you get to 1 the error free in C.

ALTERNATIVE 1

from the table of values of x and y, get the CENTROID. thats the average point((mean value of x), (mean value of y))

now plot the centroid on the graph, most probably it will lie on your best fit line.

connect the centroid with a plotted point that deviates the most from your line of best fit

you get the worst acceptable line

ALTERNATIVE 2

there are vertical error bars right? take the lowest bar of the lowest point and connect it to the highest bar of the highest point, and you get the worst acceptable line.

or you can do it the other way round too...

now error in gradient= (gradient of worst line) - (gradient of best line)
error in C=0.5(error in gradient / gradient of best line)
[yes, multiply the root power
because m=c^2,
so c=m^0.5 ]

its kind of a lot to do, but its the cie way...