I'm trying to plot the following data file
#x y s err
1 1 0.1 0.2
2 2 0.2 0.2
3 3 0.3 0.2
4 4 0.4 0.2
5 5 0.5 0.2
6 6 0.6 0.2
7 7 0.7 0.2
8 8 0.8 0.2
9 9 0.9 0.2
10 10 1.0 0.2
where the points have a variable size given by column 3 and the errors are given in column 4. I can get
plot "test" u 1:2:3 pt 7 ps variable
plot "test" u 1:2:4 w yerrorbars pt 7
to work independently, giving me this:
But when I try to combine them
plot "test" u 1:2:4:3 w yerrorbars pt 7 ps variable
I get something very strange:
yerrorbars
seems to be using column 4 as the y column and column 3 as the yerror column. Even stranger, I get the same output if I try u 1:2:3:4
. Is there something wrong with how I'm doing this? I can manually draw the errorbars as vectors, but I'd prefer to use the built-in errorbars style if possible.
gnuplot> help yerrorbars
The `yerrorbars` (or `errorbars`) style is only relevant to 2D data plots.
`yerrorbars` is like `points`, except that a vertical error bar is also drawn.
At each point (x,y), a line is drawn from (x,y-ydelta) to (x,y+ydelta) or
from (x,ylow) to (x,yhigh), depending on how many data columns are provided.
The appearance of the tic mark at the ends of the bar is controlled by
`set errorbars`.
2 columns: [implicit x] y ydelta
3 columns: x y ydelta
4 columns: x y ylow yhigh
An additional input column (4th or 5th) may be used to provide information
such as variable point color.
So in order to provide more than 3 columns and still use a single value for the ydelta, you should be able to do
plot "test" u 1:2:($2-$4):($2+$4):3 w yerrorbars pt 7 ps variable
However, as you point out this doesn't actually work as documented.
Work-around
An alternative is to make two passes; first plot the errorbar lines and suppress the points, second plot the point with the desired properties :
unset key
plot "test" u 1:2:3 with yerrorbars pt 0, \
"" u 1:2:4 with points pt 7 ps variable