It is now generally agreed that the band spectrum of helium, which was first observed by Curtis (‘Roy. Soc. Proc.,’ A, vol. 89, p. 146, 1913) and by Goldstein (‘Verh. d. Deutsch. Phys. Gesell.,’ vol. 15, p. 402, 1913), is to be attributed to some molecule of helium. This band spectrum is peculiar in the fact that the heads of the bands have been shown by Fowler (‘Rov. Soc. Proc.,’ A, vol. 91, p. 208, 1915) to follow the law usually associated with line spectra, though the individual lines composing the bands can be represented by the parabolic arrangement appropriate to band series. More recently, Curtis has carried out a series of investigations (‘Roy. Soc. Proc.’) on the structure of the bands in terms of the quantum theory. Attention may here be drawn to two peculiarities in the spectrum. There is one isolated band with a head at about λ = 5733 A., which is degraded to the violet, whilst all the remaining bands are degraded to the red. Also Goldstein ( loc. cit. ) observed a number of faint band lines in the region about λ = 5390 A. to λ = 5270 A., which were not recorded in Curtis’s paper ( loc. cit .). It is well known that in vacuum tubes excited by uncondensed discharges only faint traces of the principal band heads are visible in the positive column though the complete band spectrum appears in the negative glow. The band spectrum can be excited with much greater intrinsic brightness by using a discharge tube with a wide tube in place of the usual capillary, and exciting it by means of a discharge from an induction coil or transformer, with a condenser in parallel and a small spark gap in series with the discharge tube, the band spectrum under these conditions appearing throughout the tube. There appears to be an optimum length of spark gap and the spectrum tends to become weaker when the length is increased beyond a certain point. Curtis ( loc. cit .) has found that the band spectrum is not strongly developed at low pressures, and this condition appears to be independent of other conditions of excitation. In the present investigation we have found that under certain conditions the band spectrum can be greatly modified. It was observed that when a vacuum tube, containing pure helium, which had been made with the capillary in several sections of different bore, was excited with an uncondensed discharge the narrowest section, which was of the finest thermometer tubing that could be worked conveniently in the blowpipe, showed nothing but the line spectrum, but in the wider sections on either side the band spectrum was quite strongly developed. This seems to show that a high-current density is not an essential condition for the excitation of the band spectrum, but it was remarkable that with these tubes it appeared in the wider parts, where it would not have been seen if the capillaries had not been provided with a section of narrow bore.