When Your Signal to Noise Ratio Sucks

Alien-Microwave-mainFor several years, Australia’s Parkes Observatory has been detecting sporadic, mysterious radio bursts that appeared to emanate from deep space. Last month, it was reported that this mystery has now been solved. Apparently, the folks at the observatory would sometimes open the microwave oven door prematurely, without first turning the microwave oven off. The telescope was detecting the brief burst of microwaves that escaped the oven. Doh!

That amusing anecdote reminded me of a similar story from several years back. Physicist Riley Newman at the University of California at Irvine was trying to measure the gravitational attraction between a steel tube and a copper rod. Think about that for a moment! You feel the gravitational attraction of the Earth beneath you because the Earth is rather enormous. Imagine trying to measure the gravitational attraction between two small objects like a steel tube and a copper rod: The possible sources of error in that measurement are insane! Riley’s team had either eliminated or accounted for all of the possible sources of noise as best they could, but they still encountered a strange phenomenon that caused their measurements to go a little screwy on occasion. It mostly occurred at night, and what would happen is a chart recording that otherwise was a level line would start to climb. It would climb for an hour or so, then level off, then very slowly decline again to it’s original value.

It took Newman’s group a while to trace the source of this errant signal. As Newman describes it, one of his students was leaving the lab late at night when the sprinkler system came on. The group did some calculations and, sure enough, the mass of the water being pumped out onto the lawn around the laboratory matched was just enough to account for the strange signal.

I have my own such anecdote of spurious data, similar to Riley Newman’s, though not nearly as cool. I set up a reactor system once so that several types of data (temperature, mass, agitation rate, etc.) would be automatically recorded. I was testing the setup and noticed that the mass of the empty reactor oscillated rhythmically throughout the 24-hour day. The reactor was jacketed, and the jacket held a coolant that would ordinarily be pumped in from a nearby recirculating cooler/heater. For the test though, the pump wasn’t on and the coolant (probably glycol-water, but I don’t exactly remember) in the jacket sat “idle”. It turned out though that the fluid in the jacket wasn’t as idle as I thought. Throughout the 24-hour day, the lab temperature (also recorded by my setup) also oscillated as the building went into a power-saving mode during the night. What was happening was that, as the fluid heated, it expanded and some of it was pushed into the lines which were supported on aluminum bars and therefore not part of the recorded weight. When the fluid cooled again, it contracted, drawing more fluid back from the supported lines into the weighed reactor jacket. Double Doh!

What would lead me down this trail of mishaps in measurement? See my post from a week ago:

“NASA” Invents Steampunk EM Engine Powered by Unicornium Lulz

You see, if your data is telling you that you’ve opened up a hole in the space-time continuum, or that Einstein was wrong, or that vaccines cause autism, you really need to check your data again and maybe get your shit together. Maybe you’ve just invented warp drive. Then again, maybe you should check the water sprinklers and microwave oven before announcing your discovery to the world! Doh! 😛

This entry was posted in Science and tagged . Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s