No, your first example is not right -- per the documentation for plot, the columns of Y when no X is given are individual lines down each column and the X is 1:size(Y,2) or 1:nColumns in Y.
WHY? "Because" that's the way it is by TMW definition. It's a logical choice in Matlab because array storage is column-major (that is, increasing first by row, then by column so a given column is a set of consecutive values in memory).
See
doc plot % and friends for more details and examples
ADDENDUM:
In the second, x and y are two vectors and z is the result on those N points where N=length(x) so is also a vector of length N.
meshgrid does what the name implies--it takes the two input vectors x any y and fills out a grid of corresponding size NxN where there's a value at each.
The easiest way to see the difference is to use a vector that is short enough that the result shows easily in the command window--try
x=-1:1; y-x;
[X,Y]=meshgrid(x,y)
Study the output until you see the pattern...
