When you are creating a model, you are not going to intentionally reduce the accuracy of the models as it is pointless and time consuming to do so. For example, with the magnetic variation data, I posted thousandths where the data has an accuracy of .5 degrees. I am not going to re-work the data in Excel and re-format it just so I can post it. It is my data and you need not accept it as your own. Again, this is a tangential conversation. If you feel that data is using excessive precision you can choose to ignore it or re-format it.
You are comparing the variation used by Williams which he got from looking at isogonic lines printed on whatever chart he was using for reference. These are only marked in whole degrees and every navigator knows that these lines are approximate and that variation changes over time. You then compare those with the predictions based on a model of the magnetic field of the earth, these were not measured values, and these predictions are not guaranteed to be accurate. You are going way beyond what the data supports with your computations.
The variations were critical to establishing the flight plan. As I stated before, I believe that these values are too crude for such an attempt over water where being off by a degree or two over just 1,000 miles could put you out of visual range of such a tiny target as Howland. For most purposes the degree of accuracy was probably fine. My opinion is that flying over such distances over water it was not fine. Variations are important in that if you are unable to obtain a fix to your flight line, and you are using DR, the error in the variations will have a cumulative effect. In this case, an unacceptably high error can accumulate.
Do you have information that suggest that his magnetic variation data was correct by any measure? I would think that the models that claim 30 minutes accuracy might be a bit better than what was available in 1937. I do not claim to understand their mathematical model but I am guessing it is well thought out and models all of this historical data collected over time otherwise it would be entirely pointless to create the models for dates in the past. The point of using this data was for my own flight reconstruction so there is not much point in discussing this model data.
What mistake are you complaining about?
The assumption in the conversation would have been that FN did not re-work the flight plan and lay down a RHUMB line. While that might have been a great idea it retrospect, do you have any evidence that FN did do this? Do you have any evidence that FN re-worked the flight plan that Williams created? You have mentioned this RHUMB line previously but I have yet to see any evidence that this was indeed the case. Even though FN might have been a great navigator he was also human and perhaps distracted and or lazy, instead using the plan that Williams had created.
As to the point about the strip chart, if you simply subtract 180 from your magnetic headings in the opposite direction, this implies that you are subtracting 180 degrees from your true course heading on the return trip. Is that correct or no? You seem to suggest that this is perfectly valid, I do not believe this is the case.
While you posted that the initial leg from Lae would have a 079.4° true, I believe that Williams assumed a true course of 77.05 that was 180 degrees from the true course to Lae, 257.05. That is why the magnetic heading is 73, which is 253-180. Is that not what is written on the chart for the return flight? Are you saying that this is correct?