April 12, 19 17] 



NATURE 



»35 



recovery of a lost electron. It is to be remarked that 

 as a rule band spectra are not subject to the Zeeman, 

 Stark or Humphreys-Mohler effect ; in the exceptional 

 cases it is probable that those subject to one ot these 

 effects are subject to all. It would be of interest to 

 examine these cases with reference to the nature of 

 the molecular charge. 



Luminous vapours emitting line spectra appear, 

 in many cases at least, to be positively charged. A 

 sodium flame is attracted to the negative plate of a 

 condenser. A metallic salt introduced near the cathode 

 of a spark discharge colours the spark only in that 

 neighbourhood ; if introduced near the anode, the 

 colour flashes entirely across the spark. The most 

 promising method of verifying such conclusions ap- 

 pears to be by the study of canal or jxjsitive rays. 

 Sir Joseph Thomson, from a study of the deflections 

 produced bv magnetic and electric fields, found that, 

 with ver\' few exceptions, no molecules of either ele- 

 ments or compounds carry a negative charge, while 

 those with positive charges are common. No molecule 

 acquires more than one positive charge. The atoms 

 of but few elements are found with a negative charge, 

 but all may acquire positive charges and many may be 

 multiply charged. For example, krypton may have as 

 many as five and mercury eight jx)sitive charges. 

 Hydrogen never has more than one charge, which 

 accords with Bohr's view that it has but one detach- 

 able electron. 



Stark has reached similar conclusions from a study 

 of the spectra of canal rays. In many cases the 

 motion in the line of sight gives a Doppler effect. 

 There is an undisplaced line due to the stationary gas 

 nnd a displaced line due to the canal rays. A distinct 

 -paration between the displaced and stationary lines 

 -hows that the canal rays cannot radiate until their 

 kinetic energy reaches a threshold value, which Stark 

 first interpreted in favour of the quantum theor\', but 

 which he now believes to represent the energy neces- 

 sary' for ionisation. There may be two or even three 

 displaced lines, with sepyarations consistent with the 

 view that the luminous centres are doubly or triply 

 charged. The radiation is evidently due to collisions, 

 for a reduction of pressure in the canal ray chamber 

 causes a reduction of luminosit\". In general, all 

 series lines are subject to the Doppler effect. Fulcher 

 has shown that nitrogen canal ravs give the negative 

 pole band spectrum, with displacements, but no other 

 bands have been found to give this effect. The series 

 lines of hydrogen sihow displacements, but they are 

 not observed in the many-line soectrum except to a 

 slight extent in a few cases. Stark concludes that 

 the series lines are emitted bv positive atom ions, and 

 the lines of the secondary spectrum by neutral atoms. 

 He thus associates the compound soectrum with band 

 spectra, which he supposes to be due to neutral sys- 

 tems. It mav be remarked that Fabrv and Buisson 

 have concluded from measurements of the width of 

 lines that both ssnectra are due to emission centres 

 of atomic size. From a study of the displaced com- 

 ponents of manv elements, electronegative as well as 

 ♦ lectropositive. Stark concluded that in all cases line 

 <pectra are emitted by positively charged atoms. 

 Aluminium atom ions may have one, two, or three 

 charges, which appear in succession as the voltage 

 is increased. The same is true of argon. The red 

 soectrum is apoarently due to singly charged ions, 

 the blue or spark spectrum to multiple charges. Mer- 

 cury- may have as many as four charges, each giving 

 rise to a characteristic group of lines, all those due 

 to multiple charges being spark lines. From an ex- 

 amination of many such cases Stark concludes that 

 in general arc lines or those of the positive column 

 are due to singly charged ions, sharp spark lines to 



NO. 2476, VOL. 99] 



double charges, and (Mffuse spaHc lines to triple 

 charges. There are some apparent exceptions to this 

 classification, but in the main the evidence seems to 

 support his views, which are also consistent with the 

 results obtained by Reachenheim from the study of 

 anode rays. For the first time we are thus enabled 

 to assign a common cause for spark lines produced 

 under apparently very different conditions. They are 

 found in the spectra of disruptive discharges, of the 

 negative glow in vacuum tubes ; in the intermittent 

 or oscillating arc when rapid changes in potential 

 occur, although the maximum potential may be small ; 

 near the poles of the arc, where the anode and 

 cathode potential gradients are ste^ ; in the electric 

 furnace when the temperature is high ; in high tem- 

 perature stars, and, as found by Hemsalech and de 

 W'atteville, even in tne green cone of the Bunsen 

 flame, where chemiciil action is energetic. In all 

 these cases we might expect multiple itmisation to be 

 favoured. 



Similar conclusions regardir^ the charges of 

 emission centres may be derived from observations 

 by Stark, Child, Strutt, and others on the luminous 

 vapours from an arc between charged condenser 

 plates. The carriers of the line ^)ectra are swept out 

 of the field, while the luminous vapours giving band 

 spectra are unaffected ; or, if the lines of several 

 series are present, their intensities are modified in 

 different degrees by the electric field. Studies of the 

 oscillator)- spark by Schuster and Hemsalech, Schenck, 

 Milner, Royds, and others indicate that the spark 

 lines do not persist as long as arc lines. If the emis- 

 sion centres of the former are multiply charged this 

 is what we might expect. 



Investigations on the mechanism of the «>ark give 

 results which at first sight seem opposed to Stark's 

 theor\\ All observers agree that the luminous vapours 

 appear to be projected from the cathode, with different 

 velocities for different lines, and the tacit assumption 

 seems to have been made that they are negatively 

 charged. That metallic vapours are projected from 

 the cathode is evident from the fact of cathode dis- 

 integration, and probably the particles are initially 

 negativelv charged. We know ver\- little concerning 

 this phenomenon, but two things are almost certain — 

 that only a small fraction of the metallic particles 

 take part in the luminosity, and that these particles 

 are not negatively charged while radiating. The 

 large velocities indicated bv the curvature of the 

 streamers viev^-ed in a rotating mirror do not giv^e 

 rise to a correspcmding Doppler effect, and it seems 

 highlv probable that Hull and Royds are correct in 

 their surmise that what happens is realh* the propa- 

 gation of a condition of luminosity through vapour 

 which continuously fills the gap after the first dis- 

 charge. Electrons initially projected with a high 

 velocit\-, which diminishes as the field intensity drops 

 to zero, and producing multiply charged ions in the 

 beginning and sin^lv charged ions towards the end 

 of their course, would apparently account for all the 

 observed effects. 



While the experimental evidence seems to favour 

 the idea that lines are emitted by positively charged 

 centres, there is no a priori reason why neutral or 

 even negative ions should not emit line spectra. It is 

 quite possible that the canal ray lines which Stark 

 attributes to sing^ly charged ions may be emitted at 

 the instant of neutralisation ; but we cannot escape 

 the conclusion that spark lines at least are emitted 

 bv positive ions unless we accept the improbable view 

 that a multiple charge may be instantaneously entirely 

 neutralised. Lenard inferred from the distribution 

 of emission centres in the arc that the lines of the 

 principal, series are emitted by neutral atoms, those 



