sketch only works with when reading one analog pin.

I was hoping that someone could help me out. I'm very much a novice when it comes to arduino.
I have posted the code for my project on bitbucket for you to take a look at.

The projects purpose is to read the analog pins, which I have temperature sensors hooked up to.
I have an android app that connect to the server and gets the temperatures for each probe.
As an example of my problem, I have probes connected to analog port 0,1,2.
0 = 118F water
1 = 37F water
2 = 87F room temp.
If I only have probe 0 hooked up, i get a reading of 118 for that probe. When I leave probe 1, both, the readings of probe 0 and 1 will become the same reading (85,85)

I tested it the other day, all seemed find at the beginning, probe 0 had an accurate temp of 250, and probe 1 was 45, and probe 2 was 47. Later on in the day I noticed that probe 1 and probe 2 temps were inaccurate, they were always within a few degrees of each other. Later on, I had to disconnect all three, and when I plugged them back in they all read the same temperature.
It seems that if I have only one probe connected, it doesn't matter if its connected to pin 0,1 or 2, the temp is accurate.
but if multiple probes were connected that's when I was having issues.
Does anyone have any idea what may be going on, or what I can do to troubleshoot the cause of this issue?
let me know if my description is confusing.

You read the channels very short after each other and as the Arduino has only one ADC (multiplexed) they influence each other.

One error

double readValue;

should be

double readValue = 0;

What you should try is to read the temp sensors in the background with some delay between them (or throw away the first read after switching pins. Then when a web request comes in than you give the "cached "values in return.

Thank you.
I did fix the readValue to readValue = 0; while I was troubleshooting.

I was also thinking of "Caching" the values, but haven't gotten to do that yet.

I've read a lot about having to add the delay. Do you think a day of 200 milliseconds is enough? or should just doing a read, then recording a read, then doing another read, and throwing away the first and last work just as well?
Once again, thank you for your help.

Joe