The advertizing literature from Bulova emphasized the accuracy of the Accutron watch compared to a traditional mechanical balance wheel watch. They stated that the Accutron was accurate to one minute a month, or 99.9977% accurate, which was considerably better than most mechanical watches of the day. Bulova stated that to achieve that accuracy, the watch may need to be adjusted for the individual user, a service they offered free for new watches. Out of the box, Accutrons are set to run at -2 seconds a day dial-up, as that was found in practise to be the optimum setting for a watch that is worn on the left wrist and dial out, the most common way a watch is worn. That is also how I set Accutrons after I have serviced them, unless otherwise requested.
It is often mentioned that the daily accuracy of the Accutron, by extrapolating the minute/month accuracy, should be +/- 2 seconds a day. Can we expect this to be a valid way to look at it? Well, the answer to that is No! you can't say that, because the advertizing clearly states 1 minute per month, a length of time that takes into consideration the many factors that affect the instantaneous or daily rate of the watch. The main factors that affect rate are Positional error caused by the design of the tuning fork, Isochronism (how the battery voltage affects the rate), Temperature in the range of 0 to 60C (which mostly affects the transistor, less so the fork as it is made from a special temperature stable alloy (NiSpan-C), Altitude and exposure to external magnetic fields. Note that the temperature variation of an Accutron that is on the wrist is quite small, just a few degrees, as the body temperature stabilizes the watch temperature quite well.
So, you can see why it is not valid to state the accuracy in seconds per day, as there could be many variables the movement is exposed to over a period of 1 month, the one month period giving a good average of everything.