I noticed that also. But it appears to me that he pretty much nailed so many other details. What I think could have happened to this one detail, is that Rowland wasn’t fully in sync with Specter regarding that aspect of the question. If you look through the upper window pane though, Rowland’s description works quite well with what the simulation shows.
It is abundantly clear Rowland misunderstood the question.
When asked what the distance was between the top of the man's head and the open window Rowland is trying to estimate how far away the man is standing from the window. He has already given this estimate a couple of questions before, 3 to 5 ft.
To imagine he is talking about a 3ft space above the man's head to where the open window was, is clearly not the correct interpretation of what he is saying.
When being asked about the SN window Rowland makes the following point:
Mr. Specter: How much of that window was open?
Mr. Rowland: It was open about that far.
Mr. Specter: Indicating 2 1/2 feet?
Mr. Rowland: Two feet.
Mr. Specter: Two feet.
Mr. Rowland: Indicating 2 feet. It looked like the windows might open 3--two-thirds or three-fourths of the distance.Here is another example of how accurate some of Rowland's observations were.
He remembers the SN window was not fully opened, as opposed to the south-west windows which he described as being opened to their fullest extent. And this was indeed the case.
He estimates the SN window was opened about 2ft and that this was only two thirds or three quarters opened. So at his maximum estimation,
the window opened to it's fullest extent would be 3ft (if 2ft was two thirds open).
It therefore makes absolutely no sense that he would then say there was a 3ft gap above a man he could see from the waist up!
Rowland is talking about seeing a man from the waist up through the estimated 3 ft gap of the south-west window opened to it's fullest extent.
It is clear Rowland misunderstood the question and was trying to estimate the distance between the man's head and the open window.