D O C U M E N T 8 7 O C T O B E R 1 9 2 5 9 9 Now the interferometer. I unfortunately have never worked with it. My col- league Prof. Henriot, on the contrary, is very experienced in this area & has agreed to help me.[2] My assistant, Dr. Stahel, has also operated one.[3] Henriot has the re- quired glass plates. I think that the instrument can be assembled this way in the lab- oratory. Now to its dimensions: Michelson used a light path of 11 m. The effect then comes to 0.4 interference fringes. I plan to build a much shorter apparatus. All the sources of error at least act in prop[ortion] to the light path, partly also in prop. to the square. But therefore the light path can be reduced up to the limit imposed by the reading accuracy of the fringes. I am initially choosing a 1m light path, hence a 50cm plate distance & no multiple reflection (a balloon instrument has to be simple.) Then the effect is 0.04 fringes, the Miller effect is 0.004 fringes. Hence a precision for the result of 0.0004 suffices. Now, the reading accuracy (according to Henriot) of the black fringes (white in the original film) is 0.01 fringes. (Because I shall be photographing the lines & observing them under the dividing-machine microscope, their absolute size is insignificant.) I shall be photographing continu- ously on film running at 1 mm/sec. At each extremal position some 10 lines can be measured 10 times (approximately from second to second, –5 to +5). This yields 100 readings per minute. The series of observations spans 4 hours (from 22h to 2h). Even if some parts of the film are disturbed, one can reckon with 100 useful revo- lutions. That yields 10,000 readings one to two weeks of work and one obtains a number of observation series, each of which has the desired precision. It should be noted that the average can be taken of an arbitrary number of observation series, even when changes to the ordinal number are unknown. One only has to have the same number of readings for each revolution & have all the readings refer to the same azimuth. The point of reference will not pose a problem. I shall probably pro- ceed as follows. I will put a small spot with a fine line on the one mirror that is il- luminated from behind. The fringes should be vertical. The film runs vertically behind a horiz. slit. Onto this plane the one mirror & the virtual image of the other are projected in such a way that the spot falls onto the slit. Thus, following a semi- illuminated fringe, a fine line (black in the original) forms lengthwise throughout the entire film & at its center at the same time. It is against this line that everything is referred. You see, esteemed Professor, that quite a number of details have already been carefully considered. Much still needs to be examined, e.g., whether the heat from the two light sources interferes & other issues. Naturally, the errors one has not thought of are the worst. I would, of course, be very grateful for any criticism from you. In conclusion, some additional technical ballooning issues: