34 PROCEEDINGS OF THE AMERICAN ACADEMY. 



observations the polished surface was always covered. The first photo- 

 graph was taken with the grating in its normal position and with all tlie 

 ruled surface exposed. The result is an image of the pin hole accom- 

 panied by two sharp streamers, one vertical the other horizontal. The 

 horizontal streamer is the longer and stronger of the two. The effect is 

 shown in the Plate, Figure 1. 



If the ruled surface of the grating be protected by a screen having a 

 rectangular aperture whose diagonal is less than the breadth of the ruled 

 surface, the orientation of this aperture throws light Upon the nature of 

 the phenomenon. When the longest side of the opening is parallel to the 

 longest side of the ruled surface, the image of the point source presents 

 exactly the same appearance as when the whole grating surface is ex- 

 posed. When, however, the aperture is set askew so that its sides make 

 angles of 45° with the sides of the ruled surface, the effect on the image 

 is striking. The result is shown in Figure 2. The vertical streamer has 

 revolved through 45° and the horizontal streamer has been broken into 

 two parts. One of these parts has turned through 45° while the other 

 and longer part has remained fixed. The streamers which accompany 

 the image seem then to consist of two components, one a cross which 

 turns as the sides which bound the ruled surface turned, the other 

 a horizontal line or streamer which remains stationary. The idea 

 that the movable cross is due to diffraction through a rectangular 

 opening at once suggests itself. This theory may be further tested 

 by covering the grating surface by a screen with a circular hole. The 

 imao^e of the source so obtained showed no vertical streamer, the 

 horizontal line was still present, tliough less intense. This result in- 

 dicates that the explanation of the origin of the movable cross is the 

 correct one. 



When the point source was replaced by a slit, the effects were of an 

 exactly similar character. 



It is easy to apply the results of these experiments to the vertical 

 continuation of strong spectral lines. In order to demonstrate that the 

 vertical streamers are due to the rectangular shape of the ruled surface, 

 we have only to cover the surface with a screen whose rectangular 

 aperture is set askew with respect to the direction of the lines of the 

 grating. We obtain the effect shown in Figure 3. Here the continua- 

 tions of the strong lines have turned and broadened, though the lines 

 themselves remain straight and sharp. The dilFraction pattern from the 

 rectangular grating aperture is no longer a cross with sharp vertical 

 and horizontal arms. The cross has turned, and, since the aperture 



