The National Weather Service has recalculated. Now forecasters say that July 2010 tied with three other years for the hottest July on record for Baltimore, and the hottest single calendar month. Period.

When we posted on the July heat yesterday, we were using the web site for the NWS Baltimore-Washington Forecast Office in Sterling, Va. That listed July 1872 as the hottest July on record for the city, with an average temperature of 81.7 degrees.

July 2010 came in at 81.5, putting it in a tie with July 1995 for second place. Or so I thought.

Now, Sterling has consulted the final arbiter on such things - the National Climatic Data Center - and concluded that their own web site was wrong. The correct average temperature for July 1872 was listed by the NCDC as 81.5 degrees.

That would put this past July in a three-way tie with 1995 and 1872 for the hottest July on the books for Baltimore.

But wait. There's more. Steve Zubrick, science and operations officer at Sterling says 1949 also finished with an average temperature of 81.5 degrees, making it a four-way tie. That, despite the fact that the Sterling web site lists 1949 with an average of 81.4 degrees.

UPDATE: Steve tells me the problem appears to be differences in the protocols established for rounding temperature averages. For example, when you take the average monthly high and low for July 1949, add them and divide by two, you get 81.45 degrees. The weather service in 1949 appears to have rounded that DOWN to 81.4 degrees. Today's protocol at the NCDC would round it UP to 81.5 degrees.

Zubrick has asked the NCDC to explain past and current policies on rounding. Math teachers: here's a teachable moment.  

So, that's the story. Sterling has always said the "text file" on their web site contains "preliminary data," and that the NCDC has the last word. In the meantime, Zubrick said, "We're going to review that text file." Stay tuned.

(SUN PHOTO: Jed Kirschbaum, 1996)