Interesting Observation on Analog Input

I have been trying to hook up a ldr to my deumilanove(s) (3 of them) (C9955),(C9953),(C9978) - now i hooked the ldr up to pin14(analog0) on 2 of the above boards and it doesnt work. on 1 of the boards, pin 15 doesnt work either. now to come to a very interesting observation, i wrote the code for the input to go with pin19 and by mistake it was hooked up to pin 14 yet there was a change in values as i moved a light source in and out of the ldr. i realised my mistake and hooked it up to pin 19 and the same difference showed u, now, me being me i hooked it up to pin14 again and there was the reading again(i was monitoring the seria port all the while!) is there a possibility that there is some data leakage or something along the pins? i would love an answer to this specific bug of pin14.
Heres the code i used...
int ldr = 16;

void setup()
{
pinMode(ldr,INPUT);
Serial.begin(9600);
}

void loop()
{
Serial.println(analogRead(ldr));
}

Unconnected inputs will float a fair amount, based on many factors: radio interference, heat, signal leakage, crosstalk, gaussian interference, atmospheric conditions, phase of the moon, budget deficits and Nigerian scammer mails. Even the connected inputs will suffer some of this, but since the ADCs are low-resolution, you won't really notice.

I believe you need to use the analog input pin numbers when you use analogRead(). And you don't need to set the pin mode.

use 0-5, not the actual pin number, but the number silk screen on the PCB