A funny thing happened on the way to determining how hot 2012 has been on a global basis: temperatures changed in 1880.

We've been hearing that 2012 has been the "hottest on record." I had written earlier that those claims were based on the contiguous United States only, or 1.5% of the earth's surface. The "global temperature" in 2012 through June was only the 10^th hottest on record. In fact, every single month of 1998 was warmer than the corresponding month of 2012.

I thought I'd update that analysis to include July's and August's temperatures. To my surprise, NASA's entire temperature record, going back to January 1880, changed between NASA's June update and its August update. I could not just add two more numbers to my spreadsheet. The entire spreadsheet needed to be updated.

I knew NASA would occasionally update its estimates, even its historical estimates. I found that unsettling when I first heard about it. But I thought such re-estimates were rare, and transparent. There is absolutely no transparency here. If I had not kept a copy of the data taken off NASA's web site two months ago, I would not have known it had changed. NASA does not make available previous versions of its temperature record (to my knowledge).

NASA does summarize its "updates to analysis," but the last update it describes was in February. The data I looked at changed sometime after early July.

In short, the data that NASA makes available to the public, temperatures over the last 130 years, can change at any time, without warning and without explanation. Yes, the global temperature of January 1880 changed some time between July and September 2012.

Surprise of surprise, the change had the effect of making the long-term temperature record support conclusions of faster warming. The biggest changes were mostly pre-1963 temperatures; they were generally adjusted down. That would make the warming trend steeper, since post-1963 temperatures were adjusted slightly upward, on average. Generally, the older the data, the more adjustment.

Who knew that old data would change the most? I tend to think of old measurements as numbers you write down on paper and keep tucked away. Apparently, data can live, much like a Constitution. It is constantly adjusted, and in ways no one can fathom or explain. NASA measures with a rubber ruler. How do we know James Hansen's thumb is not on the scale?

To be fair, the overall result was that the 131-year trend now calculates to 0.64 deg C per century instead of 0.60 deg C per century. And the trend since 2002 is still a cooling one. (In fact, the cooling trend since 2002 is steeper with the new data.) So maybe this isn't that big of a deal.

But I find the idea of constantly changing data troublesome. I'm from the old school of science in which events in 2012 could not change events in 1880. Yes, I'm schooled enough in statistics and data analysis to know that estimates of past events can be adjusted based on more recent knowledge. But something is suspicious when the oldest data change the most - especially when it all comes without explanation.

But just to calm everyone down, here are some conclusions based on NASA's global "Combined Land-Surface Air and Sea-Surface Water Temperature Anomalies (Land-Ocean Temperature Index, LOTI)."

• The first 8 months of 2012 were only the 10^th hottest on record.
• Every month of 1998, except May, was warmer than the corresponding month of 2012. May 1998 and May 2012 are tied for hottest.
• While May 2012 was the hottest May ever (tied with May 1998), July was only the 13^th hottest and January was the 21^st hottest.
• The global trend since 2002 is one of cooling.

You might also be interested to know that Antarctic sea ice set another record in September: the most amount of ice ever recorded.

Once again, the basic global warming story, even after our hot summer in the US, is the same as I described in May. "In short, the data show nothing alarming at all: very mild warming over the long term, and actual cooling over the short term."

Randall Hoven can be followed on Twitter. (If you are familiar with NASA's "living temperature record," feel free to email me at randall.hoven@gmail.com.)