When you activate `plotdata`
a welcome message followed by the first `PLOTDATA:`
prompt is printed in the text window, and another window pops up, the
one in which `plotdata` will display the graphics. Only one window
can be active at any given time, so you may need to move the mouse
inside the window that you want to ``connect'' your keyboard to. You may
want to resize and/or move the text window around so that you can see
both at the same time. Do not resize or close the graphics output
window; it will get closed automatically when you exit `plotdata`.

Let's assume you have started `plotdata` successfully. At `PLOTDATA:` prompt type the following:

`PLOTDATA: x={1,2,3,4,5,6,7,8,9}
PLOTDATA: y={0.05,0.10,0.14,0.19,0.25,0.30,0.34,0.40,0.45}
PLOTDATA: graph x,y`

Your graphics window should be showing the graph of `y` *vs*. `x`!
(You can jump ahead and take a peak at Figure 1
if you are not running `plotdata`, typing these commands as you go along).

The above data could have been something like time and distance from one of your air track experiments. In fact, we should have also entered the error bars at each data point, like this:

`PLOTDATA: dy={0.02,0.07,0.01,0.04,0.05,0.10,0.02,0.04,0.04}
PLOTDATA: clear
PLOTDATA: graph x,y,dy`

We have used `clear` to clear the graph and start a new one with
the next `graph` command.

You can immediately see that the dependence is roughly linear. Let us
investigate closer. First of all, the line connecting the points is
really inappropriate, since we only made a discrete set of measurements
and really know nothing about what `y(x)` is like in between our
data points. To fix that, we will change one of the settings of
`plotdata`, that of the plotting character,
or `pchar`.

`PLOTDATA: pchar -10
PLOTDATA: clear
PLOTDATA: graph x,y,dy`

Now each data point is indicated by a triangle (the plotting character No. 10), and there is no line connecting the data points (the minus sign in front of 10).

Now we are free to use a line to indicate a theoretical model that fits this data. From the way we generated our data (air track) we expect a linear dependence. Let's fit the data to a linear expression:

`PLOTDATA: vary a
PLOTDATA: fit y=a*x`

The output generated by `plotdata`, among other things, tells us
that the best value for the parameter `a` is 0.0495+/-0.0039.
The error in the fit is given by the standard deviation (the `E2`
parameter), and there are some other useful numbers reported which we
will ignore for now.

Armed with this best fit, we can add a ``theoretical'' line to our
experimental plot. We can recalculate what the values should have been
if the experiment yielded exactly the linear dependence we expect, by
``updating'' a new vector, `f`, to contain the values that the
last `fit` command had calculated:

`PLOTDATA: update f
PLOTDATA: pchar 0
PLOTDATA: graph/noaxes x,f`

Two points of interest here: we used `pchar 0` to turn the plotting
character to none and the line through the points to on, and then we
used a switch (`/noaxes`) on the `graph` command to add the
second graph to the already existing one. Looks pretty cool.

To add the final touches, let's label things properly:

`PLOTDATA: xlabel "Time, s"
PLOTDATA: ylabel "Distance, m"
PLOTDATA: replot`

Figure 1 shows what your plot should look like at this point.

PostScript file

Before we advance on to more elaborate things you must learn how to end
your `plotdata` session. The magic word is `quit`. Try it
now.

One remark: your graph may look slightly different. In creating Figure 1 I had doubled the size of the plotting character to 2% of the page width, up from the default 1%, using the command:

`PLOTDATA: set %charsz 2`

before the `graph` command. It's one of those slightly more
advanced commands that we will get to soon.