Average of 4 standard errors in replicated plot experiment

Hello everybody!
After looking for a solution for good 2 hours, I decide just to ask for help in a nice forum.
I have a field experiment with 4 different plots. In each plot I collected and analysed 3 samples and calculated the mean and SE of each plot.
Plot 1: M1=0.2 (SE1=0.02)
Plot 2: M2, SE2
Plot 3: M3, SE3
Plot 4: M4, SE4

Anyway, now I wanted to calculate the average for the 4 plots M and its SE.
My idea is to calculate the sum of SE's, divide this by 4 and do the squareroot of it:

SE=SQRT(( (SE1^2)+(SE2^2)+(SE3^2)+(SE4^2)) /4)

Is this the right way to do it?
Or are you dividing by 4 after doing the squareroot of the sum?

SE=(SQRT((SE1^2)+(SE2^2)+(SE3^2)+(SE4^2))) /4

Or is this way completely wrong?

Thank you so much in advance! :)