Simply incorporate another string field in the format string for the observation name (and remember to include the additional space or other name for the column in the row header for alignment).
When it's a mixed array of character and numeric data, however, you have to use a loop to handle the character array element and the numeric array row elements for each row to intermix the two; if you write the two arrays as the arguments to fprintf separately, it expands each completely in order, not by index within. There's a real advantage of the "implied do" construct in Fortran for i/o
There's a new table tool in last releases as well that will automate this I believe but I don't have newer than R2012b to test....
