Floating Point Number + NAN

Code:

---

#include<stdio.h>
#include<conio.h>
float square(float);
void main()
{
    clrscr();
    float a,b;
    printf("ENter a Number");
    scanf("%f",&a);
    printf("Number=%f",a);
    b=square(a);
    printf("\nSquare of %f is %f",a,b);
  getch();
}

float square(float x)
{
    float y;
    y=x*x;
    return(y);
}

---

In the above program, I am calculating the square of float number. But sometimes the number is entered as NAN and sometimes Output is NAN. What is NAN? I am entering floating point number, then y NAN is entered?
SEE the Image attached for the OUTPUT.

NAN means Not A Number, so the conversion failed for some reason and the resulting bit pattern of the float is not a legal floating point format.
I'm not sure why it would be failing. The code looks ok to me, other than the () around the return value, but I don't know C++ well enough to say whether than matters or not. You don't seem to be getting NAN on the result, but from scanf.

What os/compiler are you using? return doesn't need () around the returned value but has no effect if present. You don't actually need to define y in square() as you can simply return the squared value. You are not checking to see if a valid number has been entered or not. Consider

Code:

---

#include <stdio.h>
#include <conio.h>

float square(float);

void main()
{
    //clrscr();
float a;

        if (printf("Enter a Number") > 0 && scanf("%f", &a))
                printf("\nSquare of %f is %f", a, square(a));

        //getch();
}

float square(float x)
{
    return x * x;
}

---