LXXXIII. The Distribution of the Active Deposit of Actinium in an Electric Field. By A. N. LuciAN, Ph.D. ( Yale) [Communicated by Prof. E. M. Wellisch.]. I. Introduction. THE observation first made by Rutherford, that the active deposit of radium is to a large extent concentrated on the cathode in an electric field, led naturally to the supposition that the deposit consisted of positively charged particles, and that under suitable conditions the concentration could be made complete. The results obtained by Wellisch and Bronson [citation redacted] and by Wellisch [citation redacted] separately showed that this supposition is not justifiable. In the second paper cited, it has been shown that the radium deposit atoms consist of neutral and positively charged particles ; that even under the most favourable conditions only a definite proportion (about 90 per cent.) of the particles is deposited on the cathode ; that there is distinct evidence of columnar or initial as well as volume recombination, and that in general both effects are much more pronounced for the case of deposit particles and gas ions than for the gas ions among themselves. The work on the actinium active deposit had been quite inconsistent from the beginning, and had given rise to various views in regard to the origin and the sign of the charges [footer] 762 [header] carried by the deposit atoms and the mechanism of the transfer in an electric field. Recently, Walmsley [citation redacted] published the results of a series of experiments in which he eliminated the disturbing causes existing in the work of previous investigators, such as presence of dust, moisture, &c, and arrived at the following conclusions : — (1) The distribution of the active deposit of actinium between the electrodes is independent of the amount of the emanation, (2) with suitable fields all of the deposit can be collected on the cathode, (3) actinium A and B acquire their positive charge in the process of their creation, (4) in an electric field actinium A and B behave exactly like the positive gas ions produced by their radiations ; they recombine in exactly the same way (initial recombination being absent) ; and the activity on the anode is due entirely to the diffusion of uncharged deposit atoms formed by recombination in the volume of the gas. The observation that the distribution of activity over a wide range is independent of the amount of emanation and that the activity on the anode is due to the diffusion of neutral particles are in agreement with the results obtained by Wellisch for the case of radium. On the other hand, some of the above results are in direct contradiction to the experimental facts established by Wellisch for the active deposit of radium, which one would expect to hold in the case of actinium also. The following investigation was undertaken with a view of distinguishing between the two contradictory sets of results and the corresponding theories, based on them, as to the origin of the charges carried by the deposit particles,, their mode of transfer and recombination in an electric field, &c. The main objects of the experiments contained in this paper may be briefly summarized as follows: — (1) Is it possible, under suitable conditions, to collect all the active deposit of actinium on the cathode ? (2) Is columnar as well as volume recombination present? (3) Does volume recombination between the positively charged actinium deposit particles and the negative ions take place at the same rate as the recombination between positive and negative gas ions ? Before describing in detail the experimental procedure, it is not out of place at this stage to draw attention to a fundamental [header] 763 difference between the method adopted by Walmsley and that followed in the present investigation. In Walmsley's experiments no direct measurement of the activity of the outer electrode (cylindrical vessel with the air-inlet on the side) was made ; the total activity was assumed to be equal to the sum of the activities obtained when the central rod was first made the cathode and then the anode. It will be shown later that this assumption is not justifiable. In the present research the activities of both the case and the central electrode were obtained by direct measurement. II. Experimental Procedure. The source of the actinium emanation employed was a preparation of actinium, which Professor Boltwood kindly placed at my disposal. The air current, which carried the emanation into the testing-vessel, was produced by means of an ordinary waterblower. This gave the desired high velocities of the air current, which could not be conveniently obtained by means of a Mariotte's bottle or a gasometer. In order to obtain constancy of flow three regulators were installed in the system : a water-pressure regulator, in series with the water-blower, and two gas-pressure regulators along with several large capacities. With these appliances the variations in the pressure were reduced to about 1 per cent. The air current was conducted through a concentrated solution of potassium hydroxide, to eliminate carbon dioxide, through two bottles of concentrated sulphuric acid to dry it, and through a train of two or more tubes containing P 2 5 , plugged at each end with glass-wool, to absorb any residual minute traces of moisture. The perfectly dry condition of the last P 2 5 tube indicated that no trace of moisture was left in the air current. This dry air current was then passed through the tube containing the actinium preparation, which was plugged at both ends with cotton-wool. Finally, the emanation was led into the testing-vessel through another cotton-wool plug. The vessel used in the experiments consisted of a brass cylinder with an insulated central electrode sprung into a split brass rod and very easily detachable from it. This rod in turn was rigidly fixed in the tapered ebonite plug P, which fitted closely into the guard-ring E. By employing sealing-wax and stop-cock grease all the joints, shown in the [footer] 764 [header] diagram, could be made sufficiently air-tight. The dimensions of the vessel are as follows: — Inner height = 25*2 cm. Inner diameter = 4*9 Diameter of inlet = 3*8 Diameter of central electrode == *151 Exposed length of central electrode... = 23*1 Distance of electrode from gauze = 1*8 [figure redacted] The diagram of connexions is shown in fig. 1. H is a [header] 765 spring-brass clip slipped over the brass rod to connect the central electrode with the electrometer system. E is a guard tube, which is always kept earthed to prevent any leakage over the surface of the insulators. F is a brass ring which was employed in order to remove the ebonite plug P and the electrode without the necessity of handling the insulation. G is a piece of fine-mesh gauze placed over the cotton-wool in order to prevent an abrupt change in the regularity of the electric field at the bottom of the vessel and also to stop cotton-wool particles being blown into the test-vessel, b is a detachable brass piece placed centrally over the gauze in order to make the flow symmetrical, as much as possible, with respect to the case and the central electrode. The vessel was supported on the side by spring-brass clips (not shown in the diagram) which were mounted on ebonite and were connected to a series of high-potential accumulators through the water resistance H and the triple key M. The wire B in the figure goes to a second vessel, of construction identical with A, except for the bottom, which was a flat brass piece in this case. KR'D is a potentiometer arrangement which was employed whenever it was necessary to charge the quadrant system to any desired potential. It might be worth mentioning that the key K was connected with the double-lever system in such a way that by closing K either way simultaneous contact was made at L with the electrometer lead ; thus the use of two separate keys was avoided. The usual electrometer key-system is represented by K 1, E, and K 2 . C is a capacity of tinfoil and mica, which when added to the electrometer system would increase the total capacity of the system 48^ times. The electrometer was of the Dolezalek type with a platinum suspension. The needle was kept charged to a potential of 120 volts ; and at this potential the sensitiveness was about 190 mm. per volt on a scale a metre away. The leads, the keys, the capacity, and the electrometer were carefully screened from all electrical disturbances. The potentials used, varying up to 1700 volts, were obtained from a battery of high-potential cells. The method of procedure in general was as follows: — The air current was first established ; the vessel was connected to an air-pump and was exhausted to a pressure of a few mms. of mercury, the outer surface being simultaneously heated moderately in order to expel traces of residual gas. The vessel was then quickly put in connexion with the air current, which was made to pass through the actinium tube ; 766 [header] the potential was applied and the air current regulated to the desired pressure. This method was followed in all the experiments, and evidently insured a perfectly dry and dust-free stream of emanation flowing into a dust- and moisture-free vessel. The electrode and the vessel were thus exposed to a steady flow for a period long enough for the emanation to get into equilibrium with its subsequent products. This period of activation was usually three hours, and never less than two and a half hours. When equilibrium was established, measurements were taken of the ionization current passing through the vessel, first with the potential (y) used in the particular experiment and then with a standard potential, which was 600 volts in ail cases. These ionization currents are designated by I v and I 60 o respectively. These operations in general disturbed the activity distribution : hence after the ionization readings ample time was allowed for this distribution to be restored. Then the actinium tube was removed, the electric field was switched off, and a strong current of dry air allowed to flow for a minute in order to drive out the emanation and deposit particles remaining in the vessel. Time was measured from the instant of the removal of the actinium tube. The amounts of deposit which settled on the case and the central electrode were measured by the ionization current to which they gave rise, with an applied potential of +200 volts on the case. Although this potential was not sufficient to afford saturation current, later experiments showed that the ratio of the activities thus measured was the same as that obtained by using higher potentials. To measure the case activity, the ebonite plug P containing the central electrode was removed, care being taken that the electrode did not touch the sides of the vessel, the gauze bottom was removed and a clean flat bottom was screwed on, to avoid spurious effects due to deposit particles collected by the cotton-wool. Then a fresh electrode held in a stiff brass rod, without the ebonite plug, was introduced into the vessel and held as usual by the brass clip H ; and a series of readings of ionization current were taken at definite intervals after the zero of time. It is to be noticed that by avoiding the use of ebonite the ionization current could be measured with great precision. As the case activities were in general small, their exact value was of relatively greater importance than those of the central electrode ; hence a greater number of readings were taken with and without the capacity C in order to check the results. It might be mentioned that almost all the readings were taken with the capacity added to the system. From these [header] 767 ionization current readings due to the activity, calculations were made o£ the maximum activity at the time when the deposit was in equilibrium with the emanation, by means of the table given in Appendix II. p. 147 of Makower and Geiger's Practical Measurements in Radioactivity. Ordinarily three or four readings were taken, and these gave for the maximum activity results which were very consistent and well within the limits of experimental errors. The activity on the central electrode was determined by removing it from its holder and suspending it in the vessel B in the same manner as described above, and measuring the ionization current to which it gave rise at definite intervals after the removal of the actinium tube. In all measurements of the ionization current, the zero of the electrometer was made the centre of the swing. This was accomplished by the use of the potentiometer arrangement mentioned above. The amount of emanation in the vessel was varied by varying the distance of the source of emanation from the vessel, keeping the pressure and hence the velocity of the air stream constant. This method was adopted, in preference to changing the stream velocity, for the reason that in this way the relative concentration of the emanation along the vessel would remain constant, as is evident from the form of [formula redacted] the expression, e u , for the concentration at any distance x from the bottom. Varying the distance of the source means simply varying the concentration at the bottom of the vessel or shifting the axis of ordinates of an exponential curve along the X-axis. The velocity of the air current, which depends upon the pressure and the resistance of the series of tubes and bottles, was roughly measured in the following way: — Exposures were made in the usual manner with different pressures, and the activities on different portions of the electrode were measured separately. This measurement was effected by inserting the wire centrally into a long brass tube of small diameter, which had a small window cut at its middle exposing about 6 mm. of the electrode, and placing the tube inside the ionization chamber of an ordinary [alpha]-ray electroscope. A large number of experiments were performed, and it was found that the distribution of activity along the wire decreased exponentially with the distance from the end ; except at the lower extremity, where there w r as an abnormal increase, obviously to be expected w T hen we consider the volume of the emanation to which the end of the electrode 768 [header] was exposed. Curves were plotted for each pressure and from these the distance x, in which the activity fell to half-value, was measured and the velocity v calculated, taking the half-value time to be 3*9 sees. It must be remembered that these determinations of velocity for a given pressure are not absolute, and that any change in the resistance of the line will change the velocity of the flow. This was noticed constantly during the course of the investigation. The pressure of the air current finally chosen was about 14'0 cm. of H 2 S0 4 . With this pressure the velocity obtained was about 5 mm. per sec, and the concentration of the emanation fell to less than one per cent, in a distance of about 15 cm., which is more than one-half the height of the vessel employed. It will be seen that with this velocity more than one-half of the emanation would be found above the first few mms. of the central electrode, the only places where any irregularities may be expected. Thus the velocity chosen gives us a fair degree of approximation to the ideal case where the emanation may be uniformly distributed in the vessel and the deposit made evenly along the length of the electrodes. As is to be expected, experiments with lower pressures, such as 9 cm. and 5 cm. of H 2 S0 4 , gave smaller values of the percentage of cathode activity, due to edge-effect and other causes which helped the case to get more than its share of the activity. Mention might be made here of one or two experiments in which it was found that the distribution of the activity was perceptibly affected (the cathode percentage activity being increased) in the case where the flow was asymmetrical with respect to the central electrode, three out of four outlet tubes being stopped. The effect of the irregularities of the field was examined in a separate series of experiments. The variation of the electric intensity in the cylindrical portion of the vessel and in the bottom will not obey the same law, no matter what the construction of the bottom is. As a consequence, we would expect to find discrepancies in the relative number of the neutral particles formed in the body of the vessel and near the bottom, and diffusing to the electrodes. The bottom corner of the vessel was filled in with a curved piece to get rid of the edge effect, and gauze bottoms of various shapes, flat and curved, tried. It was then found that, although discrepancies in the values of the cathode percentage for different shaped gauzes occurred, these never exceeded the limits of experimental error, and did not exhibit any [header] 769 consistent direction, as one would be led to expect from the consideration of the different shapes of the bottoms. Evidently the high velocity of the air stream helped to smooth out the irregularities that would be expected at lower pressures. Finally the curved gauze, shown in the diagram, was adopted in order to match up with the curvature of the corners. It will be noticed that the distance of the electrode from the gauze is adjusted so that the average electric intensity at the bottom of the vessel would not by any chance be smaller than in the body of the vessel. III. Experimental Results. Before giving the experimental results of this investigation it may be useful to recapitulate the transformations which a quantity of emanation undergoes, according to our present state of knowledge. The complete scheme of transformation, after Marsden and Perkins [citation redacted], is as follows: — [figure redacted] All the products after the emanation compose what is known as the active deposit. It will be noticed that the determining factor in the distribution of the active deposit of actinium is Act. B. Act. A being a short-lived product decays rapidly before any appreciable amount of it reaches the central electrode, for smaller potentials. In fact it may be shown by calculation, using the following expression for the fraction of the Act. A particles that reach the central electrode, [formula redacted] 770 [header] where V = applied potential, [formula redacted], mobility o£ the positive ion, assumed to sec. volt J r represent approximately the mobility of the deposit particle also, [formula redacted], the radius of the central electrode, [formula redacted], the inner radius of the vessel, [formula redacted], transformation const, of Act. A, that with 600 volts about 25 per cent, of the deposit reaches the wire as actinium A, with 1000 volts less than 40 per cent., and with the highest potential used (1700 volts) less than 55 per cent. Since, as will be shown later, no increase of potential difference above 1000 volts appreciably alters the percentage of the cathode deposit, it may be assumed that with increasing potentials the increased amount of actinium A on the central electrode has no effect on the distribution of the deposit ; at least, in so far as the final result is concerned. It seems probable, therefore, that actinium A and actinium B are born with the same physical characteristics, and exhibit the same peculiarities in an electric field. The following experimental results were obtained in connexion with the three special objects of investigation men-tioned in the introduction. First of all, the dependence of the distribution of activity on the amount of emanation employed claimed attention and was made the subject of investigation. It was found that the cathode percentage, which we shall call Ay (activity at potential V, referring to the cathode), depended to a great extent on the amount of emanation used, for values of V up to 600 volts or thereabouts. Above Y = 600, Ay was independent of the amount of emanation used, when the amount was not excessively large. The curves in fig. 2 represent the results of these experiments and give an idea of the ratio of amounts used in these experiments. Very excessive amounts have not been tried, but there are strong indications that the above independence would no longer hold good. The points on the curves represent the percentage of the total activity which is deposited on the cathode, for a given amount of emanation. As has already been mentioned, the activities on the case (anode) and the central electrode (cathode) were measured separately and from these measurements [header] 771 the above values o£ the cathode percentage calculated. A condensed table, in which the values have been interpolated [figure redacted] Diva, per second corresponding to total amount of emanation. from the curves to correspond to given amounts of emanation at points sufficiently near the experimental values, is given below, for purposes of reference : — Table I. Amounts in divs/sec Percentage Cathode Activity. [formula redacted] It is to be noticed that the curves for 600 volts, 980 volts, and 1700 volts are horizontal ; the last two being coincident. Owing to the presence of neutral deposit particles, even at the highest potentials employed, it is necessary to correct, for diffusion, all the experimental values of the distribution of activity. This was done by obtaining experimental values 772 [header] for the distribution in the absence of an external electric field (0 volts), so that diffusion alone was operative. It was found that the deposits on the central rod and the case were in the ratio of 1 to 6'6 approximately. This ratio is quite different from the ratio of the areas of the central rod and case, which was 1 to 37. It is evident that if we corrected for diffusion on the supposition that the diffusion distribution was proportional to the exposed areas, we should get higher values for the cathode percentage. All the values given in the present paper refer to distributions for which the proper correction has been made. The curve for volts in fig. 2 would be expected to remain horizontal for all amounts of the emanation employed. The upward slope of the curve for very small amounts may be due to residual electric fields as well as to molecular agitation and initial diffusion from the recoil-column which are of no effect for larger amounts. Some experiments of not a very high degree of precision were tried with a negative potential applied to the case, the central electrode being earthed. A typical set of equilibrium values with added capacity for —600 volts on the case is as follows : — Maximum activity on the central electrode ... = *08 '- sec. Maximum activity on the case = 10*7 „ Total activity = 10-78 „ Cathode percentage , . = 99*36 % Corrected for diffusion = 94 e 6 „ The corrected value for the cathode percentage is quite in accordance with the values for + 600 volts when the central electrode was made the cathode. It is now evident that if we followed Walmsley's procedure and obtained the distribution of the active deposit by measuring the activity of the central rod, first as cathode and then as anode, neglecting [A close examination of the figures given by Walmsley in Table I. (loc. cit.) seems to warrant the statement here made that he did neglect consideration of the activity which diffused to the case (anode) at high potentials.] the deposit on the case, we should obtain for large potentials a cathode activity apparently very nearly 100 per cent, of the total amount. The experiments with a negative applied potential show further that the activity deposited on the anode is due [header] 773 entirely to neutral particles ; for, if negative particles existed in the vessel, the activity when the central electrode is made the anode should be larger than the amount which settles there by diffusion alone ; whereas by a simple calculation, from the value of diffusion on the case, we find that for + 600 volts on the case and a total amount corresponding to [formula redacted]. the activity diffusing to the central electrode is *113, which is larger than the value obtained here (*08) for —600 volts on the case. This shows conclusively that no negative particles exist, or at least, take part in the transfer of activity considered. The above difference between the values of the active [figure redacted] deposit that diffuses to the central electrode as anode and cathode can be explained by considering the different relative distribution of charged deposit particles and negative ions, the different conditions of electric intensity and the consequent different amount of recombination near the central electrode and in the neighbourhood of the boundary of the vessel. The curves shown in fig. 3 are plotted from those in fig. 2 by taking the points in which the curves of fig. 2 intersect a vertical line, corresponding to a given amount of emanation. The earlier portions of the curves show clearly the increase of volume recombination with the increase of the amount of emanation employed. The gradual rise of the curves after 774 [header] 200 volts shows the presence of columnar or initial recombination. The curves of ionization coincide at about 600 volts ; the curves for the distribution of activity coincide at about 1000 volts (not shown in the fig.), and thenceforth, no increase of voltage alters perceptibly the percentage of the cathode deposit. Thus there seems to be a definite limit (94*9) to the value of the cathode percentage ; this limiting value for actinium is considerably greater than the value (88 - 2) found by Wellisch for the active deposit of radium. The dotted lines in fig. 3 show a number of curves representing the variation with voltage of the ionization current when various amounts of the emanation were in equilibrium with the deposit products. The readings of the ionization current were obtained while the air current was passing, and on this account were not of a high order of accuracy as the amount of emanation present in the vessel was subject to slight variations. Hence a large number of these readings were taken for various voltages and for different arbitrary amounts of emanation used, and a set of average curves was plotted. All the ionization curves in fig. 3 were plotted by changing the scale of ordinates to correspond to a saturation value of 94*9. It should be noticed that the ionization current is assumed to attain its saturation value at 600 volts. This is very approximately true ; at any rate the qualitative results that will be drawn from the nature of the ionization and activity curves are not invalidated by this assumption. At 1000 volts ionization, current readings showed no appreciable difference from the values at 600 volts. From an inspection of the curves of fig. 3 it will be seen that for any given amount of emanation the " activity " curve lies continually below the ionization curve. In other words, the electric field is able to bring to the central electrode a larger proportion of positive ions than of positively charged deposit particles. For smaller voltages this can be easily explained on the supposition that the deposit particles and negative ions contained in the volume of the vessel combine much more readily than the negative and positive ions among themselves, i. e., volume recombination takes place at a widely different rate for the two cases. The same remarks hold true in the case of columnar recombination also. The fact that the central electrode receives, even for the higher potentials, a smaller proportion of charged deposit particles than of positive ions shows that the deposit particles are liable to lose their charge by recombination in the columns more readily than the positive ions. The [header] 775 difference in this particular phenomenon may be more strikingly shown by considering the curves of ionization and activity corresponding to an infinitesimal amount of emanation in the vessel. These were obtained by the method used by Wellisch for the case of radium by producing the curves of fig. 2 and of: the corresponding figure for ionization currents, so as to intersect the axis of ordinates, and plotting these points of intersection against the potentials, and are marked o in fig. 3 ; they may be regarded as limiting curves, corresponding to the absence of volume recombination. They show clearly that any given potential is able to prevent columnar recombination of ions much more easily than of active deposit particles. The two curves approach at about 600 volts. IV. Summary and Discussion of Results. 1. When actinium emanation is mixed with dust-free dry air and allowed to come into equilibrium with its active deposit, the percentage of the deposit which is collected by the cathode increases with increasing potentials, but even under the most favourable conditions and at the highest potentials applied there seems to be a definite limit to the percentage of the active deposit which settles on the cathode. This limit is 94*9 per cent., or 95 per cent, roughly. 2. The remaining five per cent, of the active deposit consists of neutral particles which reach the electrodes by diffusion. It was shown also that no negatively charged deposit particles take part in the transfer of activity. 3. For values of the activity distribution which are less than this limiting value, the formation of the neutral particles is explained on the view that the deposit atoms recombine with the negative ions in the volume of the vessel for small applied potentials, and with negative ions formed in the columns for larger potentials. Thus both volume recombination and initial or columnar recombination have to be taken into consideration for a complete explanation of the experimental results. 4. It has been shown that both volume and columnar recombination take place at a greater rate between the deposit particles and ions than for the ions among themselves. This was shown by a comparison of the two sets of curves in fig. 3, one for equilibrium ionization current and the other for the cathode percentage of the equilibrium active deposit. This behaviour of the deposit atoms leads one to the 776 [header] conclusion that they are o£ larger mass and size than the ordinary gas ions. The ionization and activity curves of fig. 3, marked o, and corresponding to absence of volume recombination, afford further evidence as to the size of the deposit atoms. It will be seen on inspection that the activity and ionization curves cut the axis of ordinates at the points "36 and '66 respectively, corresponding to 38 per cent, and 69 per cent, of the total number of ions and of deposit atoms. These numbers represent, according to Wellisch and Woodrow [citation redacted], the percentage of the total number of ions and of deposit particles which escape from the [alpha]-particle and recoil columns as a result of molecular agitation and diffusion. The above numbers indicate that, roughly speaking, twice as many positive ions on the average escape from the [alpha]-particle column as positively charged recoil atoms from the recoil column. This relative slowness exhibited by the deposit particles is naturally to be ascribed to their size and mass as compared with the ions. In all these particulars actinium active deposit seems to behave, qualitatively at least, like the deposit of radium. A theory has already been advanced by Wellisch in explanation of the behaviour of the radium active deposit in an electric field. According to this view, after a deposit particle recoils into the gas, it is subject to the chances of columnar and volume recombination. But when both columnar and volume recombination are avoided by the application of sufficiently high potentials, the distribution of the active deposit on the electrodes is determined by the relative number of charged and uncharged carriers which result from the process of recoil. During the motion of recoil the deposit atom is unaffected by any applied electric field, so that initially the relative number of charged and uncharged recoil atoms is independent of the applied potential. The nature of the charges carried by the deposit particles at the end of their recoil path is determined by the continual process of gain and loss which occurs during the recoil motion of the particles. This theory is susceptible to modification and further development, especially w T ith regard to the sign of the charges acquired by the deposit particle as a result of and at the end of the recoil motion, by taking into consideration the mechanism of the process of ionization in the following manner. To start with, the recoil atoms at their formation will acquire in general an electric charge as a result of the simultaneous expulsion of an [alpha]-particle and a number [header] 777 of slow- moving electrons ; further, the deposit atoms are at least of ionic order o£ magnitude and perhaps larger, and move with relatively small velocities. We may also assume that the recoil atoms effect the process of ionization by some sort of collision with or bombardment of the gas molecules, expelling negative electrons from the molecules with much greater initial velocity than that imparted to the main bulk which forms the positive ion. We may picture to ourselves, therefore, the recoil deposit atom moving all the time in and at the head of a column the core of which consists of an overwhelming majority of positive gas ions and the outer layer is constituted of negative electrons. Now, whatever the charges acquired by the deposit particles at their formation, it is only reasonable to expect that travelling in this core, lined on all sides almost exclusively with positively charged gas ions, they should possess a greater probability of emerging from the recoil columns with a positive charge than otherwise. If the positive charge is acquired by the combination of the deposit atoms with the positive ions of the core, then the resulting positively charged deposit particle assumes molecular dimensions and becomes a cluster under favourable circumstances ; this agrees very well with the conclusions arrived at from the discussion of the curves of fig. 3. Further, the more complete the initial exclusion of the negative electrons from the columns, the less will be the number of deposit atoms emerging from the columns as neutral particles, and hence the higher the limiting value of the percentage of the cathode deposit. It is also evident that no negatively charged deposit particles, even if they existed initially in great numbers, could emerge as such from the recoil columns. The supposition here involved is that recoil atoms of different initial velocity impart different amounts of energy to the electrons which they expel in the act of ionization, and as a consequence, for a very short time, during the motion of recoil, a separation of positive and negative columns of ions actually occurs. The fact that the limiting value of the cathode deposit for actinium is higher than that for radium is exactly what would be expected when we consider that the recoil atom from actinium A is expelled with a greater velocity than that from radium A. Moreover, since the recoil deposit atom of thorium has a velocity intermediate between that of radium and actinium, we would expect to find the limiting value for the cathode percentage of the thorium active deposit to lie between SS'2 and 94*9. [footer] 778 [header] Experiments on the distribution of the active deposit of thorium are now in progress and will soon be published. In conclusion I wish to express my gratitude to Prof. E. M. Wellisch for suggesting this problem and for his continual interest and advice throughout the course of the investigation. I am also indebted to Prof. Boltwood for his helpful suggestions. Sloane Physical Laboratory, Yale University, New Haven, Conn. May 18, 1914.