Just for future reference, it would be a good idea to use code tags around your code.
Anyway, as for nested loops, if you did something like this:
This works by setting x = 0, y = 0 and then slowly incrementing them one by one. In other words, the inner loop will get executed 40 times for each time the outer loop is executed once.
for(x=0; x <= 40; x++)
for(y=0; y <= 40; y++)
if( ( (x*x) + (y*y) == 1600)&&(x < y) )
/* Do some stuff here */
I.e. Iteration 1 - x = 0; y = 0
Iteration 2 - x = 0; y = 1
Iteration 41 - x = 0; y = 40
Iteration 42 - x = 1; y = 0
Iteration 81 - x = 1; y = 40
Iteration 82 - x = 2; y = 0
And so on, so forth.
I included the x < y statement in the if statement so that you don't get opposite repeats (e.g. x = 0, y = 40 and x = 40 and y = 0 will come out once as a solution). There are alternative methods of limiting the solutions found so that you don't get repeats.
The loop you had involved the inner loop counting down. It works in almost the same way as the loop above except that y starts at 40 and counts down.
Remember though. This method of finding solutions is quite inefficient. It would probably be a better idea to try something like this:
Since that way you only have to loop through one variable, not two.
for(x = 0; x <= 40; x++)
y = sqrt(1600 - (x*x));
if( y - floor(y) == 0 )
/* Solution found */