Interesting... Can you tell me what the source data was?
The source data was the Itasca deck logs July 2-12. I took the 8am, 12noon, 8pm latitude and longitude observations and converted those to the point locations. The linear tracks in between the points are crude connect-the-dots. Here is a snippet of the Itasca deck log coordinates that were used in the conversion. One of these exists for each day.

If that is an accurate log of the Itasca movements I am a bit bothered by that Itasca radio message that stated that they had search from 320 degree to 45 degrees true from Howland out to 120 miles.
They also left the large area that described the most likely position on the water from 337 to 45 degrees true from Howland out to 250 miles largely un-searched. Am I missing something here?
Nope. Doesn't look like your missing anything. Unless, I'm missing the same thing!
I've revised the .kml file a bit. You'll notice two cones of probability 250nm and 120nm. Like you said, the 250nm cone is pretty void of any search tracks. The 120nm cone contains a reasonable amount though. I don't know what to make of that.