The cause of the emission of the negative band system of nitrogen from the upper atmosphere during twilight is investigated. A study is made of the two possible excitation mechanisms, N
2
(
X
1
Ʃ
g
+
) +
hv
→N
2
+
(
B
2
Ʃ
u
+
) +
e
and N
2
+
(
X
2
Ʃ
g
+
) +
hv
→N
2
+
(
B
2
Ʃ
u
+
). It is shown that the latter is far more effective than the former, irrespective of the assumptions adopted regarding the solar flux in the unobservable spectral region. From the transition probability associated with it (which is evaluated in the appendix) combined with various intensity estimates, an upper limit is obtained for the number of N
2
+
ions normally present in the
E
and
F
layers during twilight. It appears that N
2
+
ions form but a minute fraction of the total ion content. The significance of this in the theory of the formation of the ionized layers is discussed. The simplest interpretation is that ionization of molecular nitrogen is unimportant; and a reasonable scheme that invokes only the ionization of oxygen atoms and molecules is available. However, by introducing certain arbitrary assumptions a more elaborate interpretation is conceivable so that the view that the
E
layer arises from the action of high-energy coronal photons, which ionize all atmospheric constituents, cannot be finally rejected. Various aspects of the layers are discussed, and observational and experimental work, which might yield evidence on the ionization mechanisms operative, is suggested. It is pointed out that the remarkable rarity of N
2
+
ions proves conclusively that recombination between the charged particles present in the ionosphere cannot be the origin of the nocturnal radiation of the nitrogen band systems. On some occasions the resonance emission at twilight is of unusually high intensity. It is presumed that this is due to incident charged particles increasing the concentration of N
2
+
ions. The possible contribution that these charged particles may make to the night-sky light by direct excitation collisions is briefly examined. Sunlit aurorae (which are essentially similar to the twilight flash) are also discussed.
The emission of the negative system of nitrogen from the upper atmosphere and the significance of the twilight flash in the theory of the ionosphere
