attending sudden Changes of Illumination. 371 



at any stage a dark interval between the corona and tbe disk ; more 

 probably the inner edge of the corona was slightly within the appa- 

 rent permanent boundary of the disk. The red corona was very 

 narrow ; its greatest width appeared to be rather less than 1 mm., or 

 about one-fifth of a degree. The effect was best seen when the 

 attention was directed upon the tinfoil strip, which for a moment, 

 after the exposure, became bright red, the coronas, or red borders, of 

 the adjoining semi- disks meeting or perhaps overlapping one another. 

 The apparent temporary excess of the area of the disk above its final 

 area, as mentioned in (1), was probably due to the evanescent red 

 border.* 



It is remarkable that repeated experiments had been made with 

 this and similar apparatus for several weeks before the existence of 

 the red border was detected, even though something of the kind was 

 looked for. The difficulty is, not to see it, but to know that one sees 

 it; when once it has been perceived it becomes very conspicuous. 

 The phenomenon is beyond doubt constantly met with, and habitu- 

 ally ignored, in daily life. Since my first observation of it I have 

 many times noticed flashes of red upon the black letters of a book or 

 upon the edges of the page ; bright metallic or polished objects often 

 show it when they pass across the field of vision in consequence of a 

 movement of the eyes, and it was an accidental observation of this 

 kind that suggested the following experiment. 



Experiment II. 



(1) The zinc plate of the last experiment was taken from the box, 

 and the aperture in the plate was covered with thin paper. A 

 ground glass lamp of 8-candle power, attached to a flexible cord, was 

 put behind it, and the whole was moved rather quickly either back- 

 wards and forwards or round and round in a small circle at a 

 distance of a foot or so from the eyes. The edges of the straight or 

 circular streak of light thus formed were bordered with red. 



(2) A 16-candle power lamp was substituted for the other. The 

 red border then appeared to have a greenish-blue band inside it, 

 slightly encroaching upon the streak of light ; probably, however, it 

 was only the apparent or irradiation boundary that was thus affected, 

 not the true geometrical boundary. 



(3) The paper was removed, and the 8-candle power ground-glass 

 lamp was again placed behind the aperture. The red could now 

 no longer be seen, but the greenish-blue border remained. 



(4) When the 16-candle power lamp was used in the same way 



* The effect may be seen without the use of the spring shutter, if a black screen 

 be held before the eyes and suddenly removed, but it is more difficult to hit upon 

 the exact position of the disk. 



