The Development of a More Precise Cosmic Ray Sounding Balloon Electroscope and Some Preliminary Results Citation Haynes, Sherwood Kimball (1936) The Development of a More Precise Cosmic Ray Sounding Balloon Electroscope and Some Preliminary Results. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/fn18-1z28. https://resolver.caltech.edu/CaltechTHESIS:10152024-223942740 Abstract No abstract. Item Type: Thesis (Dissertation (Ph.D.)) Subject Keywords: (Physics) Degree Grantor: California Institute of Technology Division: Physics, Mathematics and Astronomy Major Option: Physics Thesis Availability: Public (worldwide access) Research Advisor(s): Bowen, Ira Sprague (advisor) Millikan, Robert Andrews (advisor) Thesis Committee: Unknown, Unknown Defense Date: 1 January 1936 Record Number: CaltechTHESIS:10152024-223942740 Persistent URL: https://resolver.caltech.edu/CaltechTHESIS:10152024-223942740 DOI: 10.7907/fn18-1z28 Default Usage Policy: No commercial reproduction, distribution, display or performance rights in this work are provided. ID Code: 16794 Collection: CaltechTHESIS Deposited By: Benjamin Perez Deposited On: 17 Oct 2024 19:19 Last Modified: 17 Oct 2024 19:25 Thesis Files PDF - Final Version See Usage Policy. 32MB Repository Staff Only: item control page

THE DEVELOPlclliNT OF· A MORE PRECISE COSMIC RAY SOUNDING BJLLOON WJ,li:CTROSCOPE AND SOME PfillaIMIN/1.RY RESULTS Thesis by Sherwood K,. Haynes In Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy, California Institute of Technology, Pasadena., Calif ornia 1956 Table of Contents I. Relation of Sounding Balloon Observations to the Cosmic Ray Problem II. The Source of Pot.ent.ial Pfoblem 1 4 General Considerations IIL Preparation of the Plates 10 Constr~ction of Condenser 12 Tests of Condensers 13 Leakage at Room Temperature 13 L eakage ai--id Capaci t;y at Low 'f emperatures 14 Behavior of Condensers in the Electroscopes 15 New Condensers 19 Tests of New Condensers 25 The Design and Construction of the.Elect.roscopes 26 Reeording Mechanism 29 Automatic Charging Mechanislll 50 Other Accessories 31 Constraction Details 32 IV. C&libration of the Electroscopes 38 V. Flight Technique 52 VI. Results of the Flights VII.. Summary 65 Re:f•erences 65 I. Relation of Sounding Balloon Observations to the Cosmic Ray Problem Cosmic Rays are penetrating radiations from outside the earth's atmosphere. They were discovered by their ability to discharge an electroscope more ra.p-idly at high altitudes th&n at the earth's surface. Since their discovery they have been extensively studied to determine their properties. Three types of apparatus have bean found. usef'ul in these studies: the ionization chamber which measures total intensity from all directions; Geiger-Mueller coincidence counters which give a measure of directional intensities; and Wilson Cloud Chambers, which enable one to study individual rays. One of the chief functions of the ionization chamber has been and still is the determination of the intensity of cosmic rays at all depths below the top of the earth's atmosphere at every point on the earth's surface. The variation in depth is important for a study of the absorption of the re.diation in the earth' s atmosphere, while the investi- gaticm of the geographical distribution is important in order to determine the effect of perturbing factors such as the earth's magnetic field. The means by which these studies are made vary both with geographical locaticrti and with altitude. Since a large proportion of the earth's surface is covered with water, the level of which is at a practically uniform pressure depth of 30 in. of mercury below the top of the atmosphere, it is natural that observations at a pressure of 50 in. of mercur.r should have been made by boat. by Clay1 ), 1927-50, The result was the discovery that the sea level intensity of cosmic rays at. the 2 equator is less than in the higher latitudes. This Latitude Effect was later confirmed by Compton2), Hoerlin 5 ), Millikan and Neher4), Prins5) and others, th11s showing that there are some charged particles present among the cosmic rays and that the magnetic field of the earth certainly does deflect them appreciably. Later work of Clay6) and Millikan and Neher4), resulted in independent discovery by each that the intensity f'ollO'!i'ted the aS-.fWJletl·ies of the earth's :magnetic field quite closely, giving riae to a so-called Longitude Effect. Further as yet unpublished results o.f !lillikan ~.nd Neher have completed a wor1d wide survey of gree.t accura cy which leaves very little doubt as to the intensity distrib1i-l:iion of cosmic rays at sea-level. Observations below sea level have been made by Kolhorster-7), Regener8), Milliks.n and Cameron9) and ClaylO) by means of sink:ing electroscopes in water or t a.king them into mines. The results so far have served mainly to show that some of ~e cosmic rays are extremely penetrating, since there are still some left at 250 meters of water below sea level. For observations {!,t hi gh al.tit udea the technique 'bEroo~es oome- what more complicated. Thero are at pras-e:nt four possibilities for the elevation of instruments. They are: l) i\{ountains; 2) Aeroplanes 6J;l_d Airships; 5) Manned Balloons; and 4) Unmanned or Sounding Balloons. Each of these present cerudn diffi-c ulties. Mountains are limited by their geographical locations, their relatively low altitudes, their inaccessibility, and the variation of their local radiation. Neverthe- less almost all observers have done some mountain work and much valuable know.ledge has been gain.ed from it. Aeroplo.nes offer higher al.,titucles and great freed-om of geographieal lo~on but have the disadvantages of tilting, and vi~ation as well as stieh severe eonditic:m1a as to tend to make obeerv~ri? somewhat unrell.a ble.. The preb1em ha::i been com,.. plet.ely solved by the Ne.nerll) vibrati-0~ end tilt .tree elec:troscope with automatic recording. Flights of from 20...ro,ooo ft. have been made from Co.mor.en:t L:ake, Canada; Spokane, Washington; March Field, California; Panama Canal Zone; Lima, Pe;ru;. and in the Philippine Islands.11}~12),li) Successful flights were also made in the Northern lJ .• S. by Mott-Smith :and Howel.1..14) That. the latitude et.feet increased :.c-apidlywith increas- ing altitude was di-s eovered by Compton.2) This -accurate aeroplane surv~y checked -this discovery, extended it to mu.eh higher altitudes and showed that at. S •.5 equivalent meters of v,ater below the top .o f the atmoSphere the intensity on the equator.was only •40% of that at the same level. in the nort..l.-iem United States. This aroused great eurios-ity as to the extent 0£ the tatitude Effect at the top of the atmosphere. For the measurement of the intensity of comic ray-fl at very :bigh alti tu.des ol'.ily mann.ed balloonS. with enclosed gondolas imd sounding b&l.loons with automat,i-c instrumen:ts are e,vailable. Both methods have bee~ uaed with considerable suecesJ-2),15) tl 6) ,.l'l} , 181 in high latitudes ao that 01tt' knowledge up to heights correspoliding to -0.'7 equivu~t meters of -r1ater in these :regions is good thOllcgh :not yet 'eomplete enough •. The C.Qlllpletely unknown quantity is the inten1;,ity -a t these. altitudes in equatorial re§µiillls.. It was this problem wlrl,eh Dr .. Millikan ·and hi.s eo-· workers in thi,s laboratory set about solving in 1955. Since high 4 altitude ~, manned ba.l.loon flights are very expensive and extremely difficult to make in any but the mo.st thickly settled and highly organized regions, it was fel.t that soundi.Rg bal.lQons offe;t"ed the only ·chance of success. II. 'l'lte Source of :Potential. Problem ·· Gel'l:eral Coneiderations It was almost half a century ago that sounding balloon flights were first made in order to obtain meteorological. data.. Since then they have been very extensively employed for this purpose. their use for determining the intensity 0£ ce•e rays at veey high altitudes has only been comparatively recent and limited.. the first in 1922. in 1952.. Bowen and Millikau19 ) were Regener1 7) and Bowen and Millikan18 ) made flights These were all which had been made prior to the start 0£ this problem, though Clay20) has since reported nights in Java whose results are probably wrong in view of Neheris recent as yet unpublished aeroplane flights in the Pbilippines. In none of these flights was the accuracy comparable to that obtained under the less stringent con.ditions in th.a laboratory, on aero:planes and in ma¥1Ued ballaons. There were two reasons for this: first, inaccuracies of pre.seu+e :Feedings, and second,. lack of s.eneii tivity of the electroscope itself'. It wa.s felt that the fi:rst, .c ould be improved . considerably by use of very high quality but. rather ex:pensive bellowe made by Julien F. Friez. end Soriis of Baltimore. This would n0t be :r,at'th 5 while~, however,. nor perhaps would the whol.e project be worth viliile unles~ the electroscope sensitivity could be increased.. The dif.fieu1ty _here was tht1.t in all previous flights the electroscope was fttlly charged before the take of£ anc:t thia one single charging had to last for the .1.-v:bole flight. Thus the sensitivity lla,d to be extremely low. If some means of frequent ~eeharging e.o uld be found, the sensitivity could be increased. This problem then became simply one of finding a source of high potential which was not too heavy to be carried by small balloons. The source oi' pot~n:tial should be able to deliver a.t least 300-400 volts a.no. should not weight ove:P l.00 grams. There seemed to be four possibilities. l~ Tiny storage batteries 2. A dry pile 3.. A condenser 4. A voltage Illllltiplier 5. Electrostatic generator The first was imm.ediately dJ.se:arded because of weight. The lightest battex-y of thi.s kind is l gram/volt whieh obviously does not meet the specification-a. Dr. Neher built a dry""'.pile which fu.netioned excellently at room temperature but whose voltage deerea$ed about 0.2% per degree Centigrade dr<>p in tempe.rature. In spite of the fact that Rege.'ler had succeeded fairly well in controlling the instrument temperature during flights., great uncertainty was felt about this .factor and so for the 6 tii11e being the dry-pile was e.bandoned. An addi tiom:.:.l reason for this was that it was questionable whether the dry-pile could be made light enough. The condenser looked rea.sonably promising as we shall see below, while the voltage multiplier and electrostatic generator actu.a.lly were not thought of until somewhat later. It was at this point t:ha t the author entered the problam. In order to consider the condenser possibilities a few .facts about balloon flights and electroscopes must be considered. They are: A. A flight lasts about 5 hours. B. It seemed desirable to recharge the electroscope approximately every 5 minutes. C. Tottl number of charges 40. The capacity of the charging switch and electroscope system would probably be about 3 cm.. but 5 cm. was assumed to allow a factor of safety. D. Because of the necessity of maintaining sri.turat.ion voltage the Neher electroscope system, which we planned to use, is sensitive only in tlie upper 25 or SO% of its top voltage. For really satisfactory results the top voltage should not vary by more than 1/10 of this or 3% of the total. This 3% was arbitrarily split into t wo qu.al parts, one of which was to a.llow for loss of voltage due to charging, and the other to account for loss due to leakage. Let a condenser of capacity C be charged to a voltage V0 and charge Q0 . When Qo is shared with rui uncharged condenser of capacity c•, the cha.rge Q which is left on C at a volte,ge V iB given by 7 If this is done n times For the problem at hand~::: 0.985, Ct= 5 cm. and n ~ C With regard be 1/2 % per hour. to = 40. =- 15,000 om_. leakage, the ~low-anee is seen :f'rom D above to The law for condenser leakage is given by V Putting t = 5600 see .. and ~ Vo =0.995 . RC= 7.2 x 105 seconds. But R = s £A - 4ff'd kA C. ... Sk ::: 9 .. 05 x 10 6 , \¥here j is the volume resistivity of' the dielectric in static units, k its dielectric constant, A its area,. and d ita thicknesa. ehanged to practical units, ~k If S is = 8.14 x 1ol8. It is thus obvious that the rate of voltage drop depends not at all on the dimensions o:f." the eondenser but only on the resistivity and speei.f'ic inductive 8 capacity of its dielectric ~ This treat~ent neglects surfaee leakage.• It will be shown bel ow how this was taken -Oare of• Table I gives S and k for the very best insulators and the expected :rate of' leakage without extra surface leak Table J; riaterial, k k Per ce11t leakage per hour Fused Quartz > 5 X 1ol8 5 > 2. 5 x 101 ~ ..C:::, 0 .16 Spetial Paraffin > 5 X 1o18 2 >l X 1019 ...:( 0 ..40 Ceresin ) 5 :x 1ol8 2 )l ·•,x 1,ol9 <: 0.40 Parawax i X lol.8 2 6 :x 1018 o.68 Hal'.d Rubber l Clear Miea 2 X lol.7 6 1. 2 X 1018 3.4 : Sulfur l X loJ.? 4 4 1 ,t. - - X 10 18 X 1ol1 • 10 .. 2 It is clear that only the first four of these satisfy the r e ... quirements. ltioa is the only one of the substances in the above table which seems at a1,l suitable mechanically for a condenser dielectric and it falls far short of the requirements. Just to be sure, however, Dr. Neller had already tested several eommerical Mica condensers and .found even the best t o l,eak a.bout 8% per hour. Of the remaining $Ubst ances, eeresi.n appeared to be the best possibility beeause according to Curt is2l) its surface .r esistivity is very high and is equslly as good at 90%humidity as at 50$. 9 At the .suggestion o.f Dr. Bowen the e.utb:or•s first endeaVQr was to make a dielectric.: from filter p&per boiled ill eereein. in a ve.oir.im. • meta.l. A condenser w£ts then i:lH'l.de with aluminum, foil serving as the plate e.nd again boiled in eeresin i n G.' V'&Cuum. Several. vie.re ma.de and in no case was the lealai.ge le$~ than 6% an hour. Next, in order to see whether even. eeresm itself was good enough a condenser was made by dipping fairl,y i'l.at pieces of s.heet eow)er into hot eeresin and stacking t,hem ap in staggered fashion so that the free ends coul.d oo connected together as terminals. Thia was SU.t!cessful in producing a leakage as low as 0.75% per hour. was 15 grsu o:t~ rut its capacity only 400 cm. It.s wei.ght Thia meant a total weight 500 grai:ne f o r l0,.000 cm ~hi ch : was . f a r too mu.ch. :It was necessary to reduce the thickness oi' the dieleeti~e considerably. As a result ne.a rly thG v.hole eummer of' J.9~3 was spent il'.\' eti effort to get sheet metal really flat. The effort vga:3 not $ueces$ful. Whenever the eeresin was made e.s thin as was necessary, a short circuit always developed. In discussing this diseoura&dng dif:f'i~ul.ty with Dr. Neher and Dr. Jehn Strong one dey, it vme suggested by them that possibly !'lat metal plates could be macle by evaponrting met,nl on mica. This seemed to the aut.r..or to be ~ very power.f\u ~..igge.e tion f or not only would the pls.t.es be flat but they would be fer elea11er tll:a.n it was posaible to ~e any meUtl plates.. After several month$ of reeearcb., the following teohai.que proved successful. 10 Prepa,ration of'~ Plates By the method of Wright&l) large sheets of elear miea wi "th •very few imperfections were split to a thickness of from 0.05 mm to o.os mm with great care being taken not to leave steps in the su.r.faces. The modification by strong 51) of using water as a help 1n splitting mica was not used bees.use even the pure-s t distilled water leaves a thin l.ayer of contamination when it evaporates. The plates were then examined for flatness and the undesirables eliminated,. By use of' templets and with a :needle as a scratching tool, these large sheets were marked into' rectangles of t wo sizes and· then eu:t out along these marks with a paper cutter ~vougbly scrubbed with xylen:e. The sizes of these rectangles can be seen from Fig. la and b: which are drawn to scale. Sometimes as. llla.l'ily as ten could be obtained from one sheet' of mica. It was then necessary to :fasten leads on to both $ides of eaeh plate. The leads used were 0.002• nickel wire which could be easily strai ghtened by pullinp, and then cut into ·1 inch 'lengths with· scissors. They were fs.stened at a · particular spot near the edge of ·the.• mica, l :;. plates by aeons of Aladdin Cement. This cement polymerizes on con.tin.:.. I ued heating and forms a hard smooth surface praetieal.ly· tangent to both the mica and the wire. Aladdin cement is a powder. When a quartz or glass fi.ber was moistened and (lipped , in the powder; enough powder &tuek to make a little bead when heated with a hot wire. By touching the end of each ll nickel wire to this hot bead, a little cement adhered to each ooe.• In this vva.y cl.l .leads were tipped with cement. To cement to the- mica one end of t.he wire was supported on the mica and the other on a hor- izontal needle at the same level. A hot wire was then applied until the cement melted,. ran out on the miea,. and was finally polymerized. As· a factor o.f safety two leads were fastened on each side of the mica. In order to avoid having the cement lump.s from alternate plates fdl Otl top of ene another and. so separate th~se plates too much., the points at wtdch the leads were fastened on opposite sides of the plates were displaced about l/2 an inch from each other. Mica masks were then made with an opening ~tbf.ving the shape. and dimensions of the shaded portion of Fig. la and b. The mica pl.ates were thoroughly dusted with -a clean r abbit's hair bl:'U.sh,. sandwiched between two masks, and then exposed on both sides to aluminum being evaporated by the proeeas devised by John Strong22) of this l.aboratory. plates after .t.his process. coat.. Fig l.a ood b snows the The sha,ded ar~a represents the .alunri.mlm In order that the leads might be. epldered la.ter1 the~ were bent in under the mask to prevent their being alwninized exe.e pt ,a.t their contact with the eement. !he coat wa.a me.de opaque on each side in order to insure good eon.ductiQn.. . .The l/l() inch ·wide lanes out to the ,·; ~ • ·oog~ were of course for the purpose of m1•k~n.g ¢oritact vrl th the .leads .. Ea.eh lead conneotd.-0n Wai$ tested for ~bout: 20 •ru;i;e;tifoamr;eres which is , . . • :· , . ' ·. : · '· .,. -:· ~:;·:._ ·~- '. -: ;: about t wice; the initiel .e har~.,~ebif);;~~ 1e~d. with 2 megohms in . -.- . ..... . ·: . •·: · . s:e-ries with a l.7 plate .c onden~er and 500 volts applied. . It WB.S very early found that the leads burned out in time if no gridleak was in 12 series. Now if all the leads on one of these plates be connected together the reault is the equivalent of an extreinely flat metal plate. The edg;eS were left uncoated to eliminate trouble du-e to their inevitable raggedness. Construction. of Condenser In order to :minimize surface leakage, the condensers "'et:' a:J.,;;ays made with an odd number of· plates, there being one more wide plate than narror1. The proper number of plates was selected, again thoroughly dusted With a rabbit' is hair b:ru.sh, and al.mo.s t completely dipped in ceresin e.t s temperature of 80° 0. In being dipped they were hung by spring clips- from the lead end and all but, the last 0.02.. /• was covered. ' ' They were withdrawn fairly slowly so that the ~Qess eeresin rnn off. As the bo:btom edge came out it was slightly inclined to allow the remaining ceresin to drain. . In thi_s way only a very slight excess thickness of ceresin was left at the bottom and this was whSTe one wa."llt ed it an-yw~. The layer was quite unifo;m and about o·..oooaw thick on &ach side.. The wide and narrow plates were then stacked alternately, staggered b-.r about 1/10 of an inch, with the leads coming out in opposite directions. Two clear wide mica plates wei·e put on the outside to prevent sticking and the whole clamped between two flat plates, heated to about 4.5 ° o, and squeezed enough to iron out irregularities and make the condenser hang together. While still in the clamp, the two sets 0£ lead wires were soldered together with hot. wire and needle, great care being taken not to get flux on the condenser 15 proper, and to clean all .flu:x from the leads when. through. finished condenser is shown in Fig. le.. fhe ·o ne very significant .fact about the design is that the narrow end is the insulated aide and so ·the only critical. place for . surface leakage is the line AB, Fig,. le. This line is completely covered with eeresin which has good su.rfa.e,e • properties .. Tests of Condensers Leakage at Room 'l'~peratu.re. Betr"1een loo. 1934 and June 1955 about 20 of these condensers were made and test.ea for leakage at foom tempe-r- ature. ~lone of them showed leakages of more than 1 .. 5% per hour. than 90% of them were under 1% per hour. More. The best run was on June, 20., 1935 wb~n over a. period of 9 nours a 7700 em condenser leaked only o.• 2% per hour. These teats were made at about 450 vol ts, bu.t the eo.ndensers proved, able to $tand 750 volts and -s till give SillUl le8kage. The general. testing procedure was to eh~ge the condenser and read the wltage with an electrostatic voltmet.e r which was a JJe:her electroscope system suitably mounted and calibn.ted. After some hours another reading vrould be _taken and the avera~ leakage ;Calculated.. In mo:at .o f the t:ests the condensers were in an enclosed box with a i:;iceative. However,, in the one o± two early eis-es where tests were made in the open, very littl.e dif.ferene.e v,ras noted. tater v.'hen •co-adense:rs :for the electro.seopes were being made, some trouble with humidity was encountered. explanation for this will be given later. The 14 L ~ e and Capacity: at Low Temperatures. Because all previous flights ' by this institution had .encountered low te.111:perto:.tutB,~, the behavior of the condensers at these temperatures had be be i..rivea-t;ige.ted.. Several runs were made which proved without question that the leakage was le:Ss at low temperatures than at room temperature.. The exact amount of this change was m,t measured for it appeared :from the results that the ea.pacit-1 increased with falling temperatu;re. This ea.used relatbrely large potential d;rops as the temperature fell and in some cases small. pot,eut,ial. rises as the teliiperature rose again. The change of capacity between 22° C and •-ro° C v,as 'therefore measured. The reSl.11. ting curve is shown in Fig. 2. The total increaae in capacity 'was 7. 4'1, or its capacity at 21° C. There could have been but three possible causes for this change.. They were increase in dielectric constant, thermal • contraction• and poasibJ.e mechanical changes. That the dieleetrie cen- stant of ceresin increases a$ the temperature falls is shown in u.. Retzow2a). Although Retzow's curve stops at 20° c, a reasonab1e extrapolation to 1owe1~ temperatures would indicate an increase of about l .. 5% and eertainl.y not more than 'Kt, between 20° C and -60° C. One also infers. that most of the change would occur before -200 C. T.he eoerf·ieient .o f thermal expansion of eeresin was measured between o° C and 20° C and £eund to be = 2.3 x io-4 per Centigrade degree. This would gtve a capacity' increase of about l ..,6% between 20° C and -50° O•. Thus only 5-4% or about htlf of the 7.4% capacity change has been a.ceounted for by the .first two !'actors. At Bowen's suggestion a con- denser was su.bJected to pressure by mean_ s of two met.al pl.at~ padded • with l/8 inch o:.f oork and clamped together. This a~com.plished two thing~ fir$t, it increased the capacity about 6% in such a. way t:h_a.t· when the pres.sure was released the condenser returned to vii thin · 1% of' it$ original capacity~ thus indicating tllat the pressure was taking up mechanic.al slack; se~owi, it reduced: the_capacity change from 7.4%, to 5%. This curve is also shown in Fig. 2. This increase in C'!.i.pacity _was annoying for it me.a nt tha.t much more apparent leakage if the temperature went,_down very much during a. This was one reaoon why an attempt was made to follow Regener' a flight. procedure and prevent, the temperature from falling very li.,.w. Behavior g! Condenaer.s B!. Electroscopee Eleven conden$ers were made for the five electroscopes. J:n order to elimina:te mee~"lical diffic:ul·ties, they were pressoo. between 0,.08ltt Dow i\i~ta1 plates with l/S2tt ru.bber pads to distnhu.i-e th-e pressure.. Light copper w:ire'.s were used to connect the insulated- side tQ an amber binding post &nd to the ,lead into the electroscope-. :turin.g the .c ali.:brat,ion.s- and tests the condensers behaved: reasona:bl.y well,.. PJ..though most o:f -the calibrations were run without e-Q-ndens-e-rs, 30 films were taken ·w ith them. .f'unctioning... Table !I shows the nUlllber of. .fili:ns for each eleetroacope on which the conden$er discharged a eertain per, cent of the .distance across the film. in S l/2 hours. Table I I Elect Ne Bf Films No of Films No of Films No of Films No of Fil.ms Voltage No over 100% 40-00% 80-100% 0- 20% 20- 40% 0 5 4 0 l 0 440 l 2 2 l l 0 740 2 4 l 1 0 0 620 5 2 2 0 0 0 sao 4 5 l 0 0 0 440 16 lO 2 2 0 Since 40% of the distance across the film represents not more , than 12% of the total voltage, we see that 87% o.f the runs gave less than 5 l/';!J; per hour, which, whi.l.e not as good as hoped _for, was cons i dered satisfactory. The f our which were wo~se were .a ll early films tnken before the condensers had become really dried out. These leakage figures also include loss of charge to the charging switch. On the flights a.t Fort Sam Houston many things went wrong. Probably the worst failure was that of the eondensers. As soon a.s San Ant.Qnio was reached the leakage on all condensers increased somewhat and that of the 740 volt and 620 volt condemter s so much that they were w.orthless . new condensers were tried bu.t were no better. These two electro-scopes were therefore used at 10:w pres-sure without condensers. Electroscope No .• 4 was the only one to be returned while Dr •. Neher was 17 at San Antonio and its condenser had functioned excellently for the hour befGre its eloek stopped. No. · O and No. 4 on its second flight · each ShQWed a rapid increase of leakage as soon as th~ flight started so that the ele:etroscope trace left · the .film after an hour •te · an hour and a half. No. S's record was so faint that it didn't indicate how the condenser worked. Upon their return, the condensers of No. O, No. 5, and No. 4's seoond flight were all short eircuited. In order to e~plain these diffieul:ties and to make the condensers foolproof, further research was carried on in the school year 1955-56. The facts to be explained before proceeding with new eondensers were, 1. The change from the early condensers which were in.senaitiYe to humidity to the later ones which were very sensitive .. 2. Failure of high voltage condensers before the others. i. The rapid in~rease of leakage im.11ediately after the ia.ke-off for a flight.. 4. fhe bad ecmdition o f the condensers upon return o f the electroscopes. Number one was explained by the discovery that there it3 more than one kind of ceresino The product which has good surface qualities is dist.illed from the natural mineral ozokerite and has a single melt,.. i n : g region of 6 5 - 7 0 ° C . The author was informed by D. D_.. faylo1• :t hat there was an .a rtificial ceresin on the market which had extremely poor surface properties and a double melting point.. It turned out that bef'Ore making the electroscope eonde.n sers a new stock of ceresin. had 18 be~n ordered from Brautt Corporation and theartifit:ia.l produet. in,$tead of the natural was :re:ceiyed. _ Thex:ef'ore gre8;,t difficulty was en~t--, e'r ed with humidity. riumbers 2, 3 and 4 are ill explainable on ~ com- bination ti>f a .~erease in the resistivity of this artific.ial wax wit:h temperatur;e or :a gradual. decreaseo£ plate separation due to the softness of the wax ,and the ,µigh attraetiGm. between the charg~ plates, or both. S~ee the lower of the t wo melting points of the a:r-tific.1.al wax was a.bout ro0 c,. end 'the attraction between the plates ve,.ried f'rom 75 gmwt/er at 400 volts to 2.50 gmwt/c'Iff' at 700 volts., it :was indeed surprising that any of the condensers worked at all at the 55° , ... 40° Q. temperatures eneounterecl at San Antonio,. Natura.lly the high voltage condensers went bad .first due to the greater attraction between the plates. As v..-:i.11 be shown later, on most ef the flights the temperature increased, slightly for the first half hour., Either because of addition- al softening due to this increase in temperature or because 0£ some phenomenon connected with the decrease of e.ir pressure around the condenser.. the plates tendecl to approaeh each other in the early portion of the flights o.f. Nos., O and 4 (second fii~t). This increased the capacity of the eondenser.s and caused a very rapid apparent leak • .That the condensers were tot.illy bad when the electroscopes were returned was du.e to an increase in this phenomenon as the electroscope .lay on the ground in the sun. That the eoruienser of Na. 4 was not rlli.ned on its first flight was due to the fact that the electroscope: was found on the day of the flight. It was found al.so that the rubber pressure pads in addition l9 to losing their quality at tbese ·high tempera.ture,s.,. al.so · ab.so.r bed eeresin. Moreover, the copper wi:re whieh was painted with this wax · dieintegrated in time.. This .may have been due to vapors from the lacquer with which the interior of the recording che.mber was painted.• New Condensers . Although natural. cereein might have come through the flights in good shape., it was decided that some mechanically strong spaeer . should be used to prevent all possibility of capacity_change,. I~ order to completely eliminate .h:umitii ty e.ffects. and ..effects due to de- creasing air pre!1l!sure it seemed necessa:cy to dip the final product in eeresin in a vacuum. Mica strips, quartz fibers, and ground fused _ quartz were all tried as spacer~. In no ease was it poe·a ible to dip the condenser in hot ceresin without produ~inga short circuit in it. This led Bow.en to suggest that mica and spaced eeresin be tried in series. Thus even if eontaet was me.de at one or two points threugh the eere.d n, ~e miea would prevent shorting and the leaka,ge through the mica would be small because of the smul area of conta<;t. Thia of CQUl'JSe ~quires that each pieee of mica be coated on only one side. ..A large nu.mber of pl.ates coated cm both sides were tl1,9eady made, so the leads and 1&11-es w:e.re merely scratched o.ff one side, leaving the rect~"'": • gular alu.ud.num coat isolated.. The condenser was made by dipping ;t he • pl.ates in cerel:lin the conventional way and ten 40 u quartz fibers were used, as spacers between eaeh pair of pl.ates • . The whole wa.s clamped between the usual Do~ Me:t ~l plates, _soldered up and dipped in ceresin 20 a.fter 12 hours in a vacuum. The resulting condenser leaked only O-.. S% per hoU:r but showed three times the capacity that was expected,. Since earesin has a dielectric constant of a.boat 2 and. mica about 6, it looked very mucb as if short circuits had as usual occurred between most of the plates, but due to the isolated aluminum coat on the back side of the mica, as soon as a short·developed th.rough a layer of eeresin, that layer was completely eliminated as a dielectric. this a plain mica condenser l'J'&s made. hour. To test Its leakage :was but 0,.5% per Thus there is a ~rising discrepancy between the 8% per hour of eowoorieal mica condensers, the 5.,4% per hour calculated limit for mica and the 0.5% per hour obtained here. Several of these condensers have been made since a..11.d have been put throu.gll quite severe tests .. Since they have :proved their wo~th, a f_airly detailed description of their construction is given here. Inasmuch as mica is the dielectric, the exact kind becomes extremely important. The mica used in these condensers was purchased • from Ellgene Munsell Company, 200 Va.riek Street, New York City. The era.er read; 1 lb. Mica, uncut, transparent, and as clear as possible, in sheets that will cut 4 inches square. 6 inches square wa~ also used. Mica in sheets that will cu.t A recea:t: lette.r from Munsell Company indicates that this was a high quality of Ruby Brazilian Mica. Since good condensers have been made from two different lots o-f this mica., ordered a yea:r apart, it is assumed that they ean be reproduced.. The ceresin u.sed is obtained from J. T. Baker Chemical Company of Phillipsburg,, New Jersey. There are two brands which have good surface qualities and single m$1.ting points,. They a.re,: Ceresin W.tue; White" a domestic' product and !iatu.ral Ceresin Wax, which. is imported f:rora Austria.' The second kind is now being used. The mica · for these cond~nsers wa,s: s;,Ut as before to e. th.ict- , ness of 0.05 to 0 .. 05 mni. It was marked out and cut .i n the seme way into the Sal!le sized plates. The wires were again 0~002 in, nickel but were now .fastened to only one side of the mica. ~..nd De Khot.ih.sky eem~nt v:m.s used inst~ of JJ.addi,n.. '.!?he rea,son for changi ng eement vras that with onl.y on:e si.tt:e o! the miea coated there were only half as many leads. Moreover, there were only a third as many plates per con.;.. denser dq.e to the higher dielectric constant. of mi.ea. hact to stand six times as much eurreJ1t as before.. Thus ea:eh lead The De Khoti11sky • proved to be mu.eh better than the. Aladdin Cement for this. •O f Aa a factor safety the protective resistance was raised from 2 megohms to 20 megohms. After the uau:al dusting proeess one side of the mies W'e.S coated with aluminum. · In o;rder to better control the thickness of the coat on trie lead connections mol'e evaporating filaments were used and only three plates were coated by each filament, as a.lo.own in Fig. 58.,. Each lead connection was tested with 500 v:olts with ::1) meg~bms in s.e ries which allows a factor of safety of 10 on normal use. .An odd nuniper of plate:s were again selected• dusted, a.nd stacked exaetly aa before with ea.re being taken that t1le eoated sides be aJJ. eithel' face down or faee up. Two alear miea plates were put on each $!de Qf the C(}ndenser and the w»le clamped reasonably firmly between the Dow Metal plates by which it was to be mGunted in the electro.scope. 'rhe leads were tied together with 0.002 in. nickel wire, and soldered together .. The insulated side 11as eonneoted by larger nickel wir$ to the &lllber plug (A), Fig. :3b. About 4 inches of this wire (D) was left free for later connection to the charging switch of the electroscope ... The uninsulated side v."as connected through the• 20 megohm gridl.ea.k C to the Dow Mewl. plates. dipping.. Fig. Sb shows the eonden.ser all ready fp;r '£he two cylindrical. plugs B are the meMs by which th-e can-- denser is to be fastened to the electroscope. The condenser was· t:..hen hMg from the boam of the dipping mecbamism ahown in Fig .. 4 l;Uld the 1 whole enclosed and evacua,ted to about 10- mm Hg for 12 or more hours.. The ceresin was· then heated to 110° C b<.r an electric heater and the e-ondenser immersed therein for from 15 minutes to ha.lf an hour. After this the condenser was raised and both eeresin and condenser were all-owed to cool, the latter oooli..11g 1m1ch more rapidly. When the cere~in was between 80° and 100 C the condenser was dipped in and i.mmediately out again several times until covered with a uniform c.oat of ceresin at least l/16tt thick. As soon as the condenser was cold, air was let in and the finished proch.1ct removed from the boom. It f\hould be mentioned metal tha~ allAparts were thoroughly wiped with :xylene before the condenser was put together. All solder :nux was also caref'ully wiped .off.. '!'.he. bottom of the amber plug cu,1d one or t wo other plac.e s including the s. s. White gridl.eak bubbled furiously in the hot cere.~ in., The condensers eometime·s bubbled a littl.e but soon stopped. It was considered advisable to keep the condenser immersed until all but the few chronic portions stopped bubbling .. Several condensers made this way were dipped in different substances. The w.bstanees were domestic ceresin, Austrian ceresirii artificial eeresin, se&ling wax, and Victron~ even at high humidities. were no good at all. The first two- were good The thil--d was good when dry. The last two They were only tried in an effort to find a really high melting point substanee which would work. One condenseP with ceresin spaced with que.rtz fibers in series with mi.ca was also • made, as well as one in which aluminum foil was used instead of evapor..- Tests of varying number and severity were tried on a.ted aluminum. several of these. They are enumerated in Table III. Table III Domestic Austrian Austrian Quartz s.paeed Sealing wax Ceresin Ceresin Ceresin Austrian Aluminum Foil Aluminum Ceresin Time to reach Fo-il 2 hrs 1% per hr after 8 hrs 6 hrs f'irst change Perc-ent leakage per hr at 25oC 0.5 60% hwnidity 0.1 0 .. 2 0.4 1.2 Percent lealcage per hr at 45°c 5. 5 20% hwnidity Percent leakage 0.6 25 c, 70% hum. 0.7 l.6 )5 0.2 0.5 - Percent leakage 45°c, 70% hum .. 0.7 1.6 1400 cm 1050 cm 0.998 1.022 Capacity per Pl.ate 1400 cm CaEacit;z 45°c Capacity 23-00 Breakdown Volta,ge >540 >1250 >540 600 cm )540 1050 cm >s40 24 High le~ages at 45° C eliminate all but the Austrian cere... sin condenser and possibly the quartz spaced one whose Aladdin cement leads burned out before this test could be made. The only- real advantage of the mica. and qua.rtz spaced ceresin condenser was its small soak in. This is altogether offset by the :bremendous amount of work to make this kind of condenser as compared with the straight mica. In justice to the domestic ceresin it should be said that this test was ma.de on the last remnants of e. 2 1/2 year old supply and that the fresh dome~tic product probably would have fared better. Since the mica condenser dipped in AustriM ceresin performed so well; and iinee t,he time was very limited, it was decided not to completely finish the investigation of quartz-spaced condensers and one or two other possibilities but to see if the excellent qualities of this con-. denser could be duplicated in others. enough fqr the electroscopes. If so they were certainly good The small change of capacity of these condensers between 23 and 45o is significant in that it means that 110 apparent leaks due to capacity change with temperature should be expected. This effect and the possible effect of a change or soak in with temperature have been f~ther tested by running the temperature up and down at abouts° C per hour with no disastrous effects. The soak in seems to be slightly lower at high temperatures. In order to test whether this condenser could be duplicated, three more were made and tested simultaneously with it for leakage at s 0 c, 22.s0 c, 57°C and 47°c. The results of' these tests are shown in Table III A. smallness of these leakages is given .An idea of the 25 by the voltage drops i n the 44 hour run a.t 22°0. They were respect- ively 14.6, 11.9, 14 . l, and 11. 5 volts out of an average of 500 volts . A contact failure pr evented the leakage of No . 2 at 47°c frOD) being obtained . The value of l . 7%/ hr . was obtained befofe it wa.s thoroughly soaked in.. Table III A. Va].ues !!1 per cent ~ _of voltage g e r ~ No . 0 No . l No. 2 No . 5 Length o.f run 5°c 0 . 01 0. 01 0.02 0.01 145 hours 22 . 5°0 0 . 05 0.07 0. 06 0.05 44 hours 27°G 0 . 22 0. 27 0 . 26 0.19 59 hours •1°0 0.7 0. 8 ( l. 7 0.8 5. 5 hours Al though Table I!I A Beems to accentuate the discrepancy l)e...tween the leakage predicted by Curtis• value for mica and the les-Jcage measured, this is not necessarily -the case. All these later condensers were coated with alwninu.m on only one side so that a portion of the dielectric may have been a thin layer of ceresin or of partial vacuum between the mica on 't.he other aide and the next coat of aluminum. • The lowest leakage obt ained with mica coated on both sides was 0.5% per hour. Since this, however, represents a discrepancy of a fac tor of 7 i t might be well to consider briefly its possible ca.uses. Curtis2l) , who made al l but @ne of the resistivity measurements quot ed, seems to have taken a great deil of care both in the preparation and environment of his specimens and in. bi s measurements. Errors due t o 26 surface leakage were supposedly completely eliminated by use of a guard ring. However, he states himself that very few of the mate~ials were so uniform that two samples would give values of the resistivity a-s close as 10 per cent, while a. factor of 10 or ev-en more was n'ot .i n.... frequent. Thus a variation in the mica itself' might account fot the discrepancy. Another possible cause might be his failure to soak the mica in long enough although he says he watched this carefully. might also be that Ohms law fails for these thin sheets. It Still another possibility is that the small leaks measured here were due to the feeding back of soaked in charge. however. The author is inclined to doubt this, It seems that more accurate measurements of the resistivity of mica should be made. It is the opinion of the writer that the problem of £-t source of' high potential f'or sounding balloon electroscopes has been satia- f actorily solved. The actual test of the solution of course must await more flights. III. The Design and Construction of the Electroscopes f/hen 1n August 1934 it was finally decided that the original ceresin condensers would be satisfactory, the problem of designing the electroscopes was begun. Since the Neher electroscopes had been so suecessful, it was decided to attempt io adapt them to balloon work rather than to follow the design of previous balloon electroscopes too closely. However, it seemed likely that the 4 inch cylindrical ionize.- 27 tion chambers from Bowen and Millikan ts 1952 flight,s could he con... verted to the Neher type without too much trouble. The first problem to be solved was that of the position of the condenser and the construction and position of ·the charging svd -t~h~ It was decided to have an au.,ulie.ry chamber above the ion chamber connected to it as far as gas flow was concerned but electrically shielded therefrom. The charging switch was a toroid wound with ·Hire with a short section cut out and mounted ns an ar-matufe. One side of this arma:l:iure was in the ion chamb~r e.nd the other in the a.tPdliary chamber. An insulated wire ran through the armature at right angles to its axis of rotation thus making contact possible betviteen the electro- scope system and the condenser. The whole switch was gravitationally balanced so that very little power vms required to run it. The conden- ser was then in a. perfectly dry chamber and .fu..nctioning under ·the conditions for 1vhich it h1:<.s been tested. Contact for charging the condenser from outside was made by means of a. gra.vi ty svd tch. The recording mechanism including clock, film drum, slit, b!l;rometer, and t.11ermometer was fastened in somewhat makeshii't fashion ~ the outside of the ehamber. Because of the necessity for more power, it was decided to use small clocks rather than watches which were used in all previous inst:mment.s. Tl.le clock used in this electro- scope was taken f1•om a cheap fad clock sold by Sears a.."ld Roebuck. The actual works were ma.de by the Lux Clock Manufacturing Co1npany of Waterbury, Connecticut. The remaining mechanisms were identical with those used on the fi nal design and so will not be <i$: $ausired, further here. The electroscope .. took much longer to make than had been anticipated and v,as not .finil!lhed until January 1935~ Condensers and electroscope system were installed and a £ew films .run. :respects the electroscope behaved quite well . In most The photographic reeord-- ing was good and the condenser functioned se well that it was hard to tell tha t it discharged at all in 3 1/2 hours. There were, however., sever&]. difficult ies, namely: l. It was too expensive to construct and not very adaptable to mass production. 2. Calibration would Qe somewhat difficult due to lack of symmetry. :5. The clock proved to be quite irregul ar, having fluctuations of 5 - 10% over periods of' t wo minutes. 4. It was difficult t0. load in the dark. Dr. Bowen, Dr. Neher, Mr. Pearson and the author drew many new designs. and had many discussions as to possibl e changes or even ra.dica.lly new designs. 'l'he chief difficulty w.=1.s that i f the condenser were to be in a chamber connected to the ion ehamber for gas f low but eleetrieaJ.ly shielded from it, there must, be pr ae:tie.ally three separate chambers provided: t,he ion ch&mber, t..tie comtenser ohambe:r,. and the recording _chamber. It was a:t this time that one of t he condensers waa tried to sea i f 1 t would work at 60 - 70% hu.midi ty. Sinee it worked 29 well, it was decided to put the condenser in the recording chamber end lead the high potential. i nto the ion chamber through a pyrex or quartz plug. This reduced the numbet· of chambers to t wo and soon Dr. lieher evol.ved a design which met objections land 2 completely and in addition promised to reduce the weight of' the instrument a little. • The elee·ti·oscope which was f inally built i s shown · diagr arrimatically in Fig. 5 and photographically in Fig. 6. It will oo desir-able to discuss f irst the general mode of operation and later the experimental details of construction. Recording Mechanism. Light, usu.ally daylight, enters the ionization chamber (Z) Fig. 5 by the window (A), pssses through the lens (B), out through the window (C), is renect.e d by a mirror (D) and exposes the film through the slit (E). The moving arm of the Neher electroscope system (F) which is mounted on a quartz rod (W) casts a shadow beam ( G) which is focused by the 1ens on the slit (E) • Both the baromet.er bellows (H) and the bi-metallic thermometer (I) have arms which lead to the slit and which are tipped with 0.0015 in .. tungsten.. These fine wires elso east a shadow on the slit so that the film has simultaneously . shadows due to the electroscope system, the barometer. and the thermometer. It was necessar.r to have a fiducial mark as well. This wa s a fine tungsten wire mo1mted on a brass arm leading from the fixed end of the barometer. The cylindrical drum {J) on the outside of which is wrapped 12 1/2 inch.es of either S. S. Panchromatic or Ple:nichrome film is rotated once around by the clock in S 1/2 hours., thus recordi.,ng the three movi1.1g shadow t:raees tmd the f'iducial trace. After this one· rotation it is autoillatica.lly stopped and the record is complete. The iiru.m and cloek are enclosed in a closely fitting spun aluminum pi1l •box which protects the film in the event: of the outer case being opened or broken. Fig. 16 shows a barometer ealibration film G1 and the flight films of lfo. 0 a11d No. 4 (1st flight). Fig. 17 shows No. 4 (2nd flight) and No. 1. Automatic Cha1·ging Mechanism. If one is to frequently recharge the el.ec:troseope :system he :inuat have some kind of an automatic charging switch. The source of high potential is of course t.he condenser {I), which is charged to 500 - 750 V'olts, depending on the partieul.a.r electroscopic system. It is connected to the amber post (V) and also to a tungsten wire ,.,,which go.es through a pyrex plug into the device (R} -, in the ionization chamber. There is a switch on the clock which connects the solenoid (N) to the flashlight batteries (M) every four minutes. Wb.en this happens the iron rod ( 0) }is pulled i1ito the sole-- noid against a. conical spring. The motion of this rod causes the small brass block (P) to rotate through an angle of about 45 degre:e«J. A 5 in. phospho;e bronze rod {Q) fastened rigidly to this block but insulated from it by quartz tubing S't,ings out through this same angle and makes contact with the ~J.ectroscope system. Simultaneously the • other end of (Q) pushes in the disc {S)· and through it makes cont,aet with the high potential tui1gsten wire. 'l'he disc (S) i,S of ambe? with a button of platinum _passing through its center. Thus when it is pressed in a s in the diagram, the platinum is insulated from the walls of the ion chamber, -vrhereas when it is out against the opening in (R) it is insulated from the high potential and connected to the walls of' the ion chamber, in this way screening the chamber from the high poten- • tial during operat.ion. P.fter one or t wo seconds of con.tact the clock switch breaks t.h f:.1 current, and the conical spring causes the arm (Q) to swing back to where it is grounded to the case. Meanwhile a. little spiral spring pushes the bu.tt,on S into its screening position. The r ate of discharge of the electroscope is then continuously recorded for 4 mi.."11.rtes until it i s a gain recharged. Other Accessories. (K) is a piece of bra ss mounted on a pivot, in such a. way that it is g.r:a.vi ta.tionally 1..mbr.i.le.nced. When the ionization chamber is below the recordiilg chamber gravitationally s peaking, the arm of this pie-ea of brass rests against the balance wheel of the clock and stops the clock. When the ion cht.,.mb..":lr is above, the ru.-m moves away from the balance wheel~ allowing the clock to turn. 'l'his furnishes a very convenient way of keeping the clock stopped until the balloon is ready to take. of'f. (U) is an opening irtto which a mirror and eye.... piece can be inserted to ascertain whether the instruJilent is working ,properly. ·The condenser can also be. ehnrged at the post V through this opening. (T) is a cha.moor in which t.Jie siec~.tive is placed. (Y) is the needle '\~ive by ·wiu.ch t.he ion chamber ca..Yl be evacuated or filled w'it ith gt;,s. 52 Construction Details. The ionization chamber is cOJnposed of two hollow spWl. steel hemispheres about o.018 in. thick and of 6 in_. dia- meter which were silver soldered together at their equators. The window holders, valve seat, gasket seat, and the truncated hemisphere of the recording cllrunber were also silver soldered on. This •sas accomplished all. at once by a process devised by Neher. All the spun parts were heated to ooo0 c in a hydrogen.furn.ace, thus being de-oxidized and annealed. The t wo hemispheres were -then fastened together by three pins at their equators. A steel plug with a hole and set screw 1n it was put over each window holder. A tlltlgsten wire spring stretched between the t wo held the windows in plaee. The valve and gasket seals were held in the same fashion by tWl.gsten v>1ire springs which hooked onto the middle of the window spring. A bar vii th hole and set screw was set across the equatorial plane of the truncated hemisphere. By means of this and another t.ungsten spring this hemisphere was held in place. Silver solder in suitable forms and &l?lOunts and paste flux / were placed between all the parts to be soldered and the whole put in the hydrogen furnace and heated to 900°C. When removed, all joints were well soldered in about hal:f the cases. In the other half the process was repeated or repairs by hand were made. This gave a. silver soldering job without oxidation or deformation. The other half of the recording chamber was a hemisphere with a flange, which was fastened by machine screws to the equatorial flange of the truncated hemisphere. Both of these hollow hemispheres were but 0.01 in. thick. Another addition to the general fra.mev.-ork of the instrument was a Dow Metal cylinder which was fa.stened by screws to the truncated hemisphere right next to the ionization cha.m,;.. ber. This in turn held on its other end a circular Dow Metal plate 0.081 in. thick in the plane of the equator of the truncated hemisphere. On this plate practically the whole recording mechanism. was mounted. The windows (A){C) were slightly tapered glass cylinders, ground into place and fastened with red sealing wax. The le11s {B) was a 7 .5 mm focal length, 7 mm diameter cemented achromatic microscope objective purchased from the Wollensa.ek Optical Company, Rochester, New York. I't; was crimped into th.:e end of a hollow Dow Metal cylinder which was threaded on the outside. On the end of this cylinder was machined a. flange to keep stray light from passing the l...ens on the outside. The lens holder was screwed and clamped into a Dow Metal arm which eame down from the cover of t.rie ionization chamber,. in or oat t.he focus could be adjusted. By screwing the lens On another branch o.f this same. arm was clamped the quartz rod (W) which he.ld the electroscope system which will be discussed later. The cover to the ionization chamber was a thin s-teel conical ~eU with a_ copper tu.be soldered on at its apex. A flange around the bQt~m of the cone was provided with a tongue which fit~ the groove of the gasket seat nicely .By means of this .flange the cover could be- bolted en with l8 #5 screws. Around the outside of the eopper tube wa-s wound a solenoid (N) whose, magnetic field, practically unaffee:ted 34 by the copper, furnished. the pull on the iron rod within the tube. The whole charging mechanism described above was mounted from thia cover so that lens, electroscope B"Jstem and charging mechanism could all be adjusted outside with assurance that they would be the same inside. In order that the mirror (D) and slit (E) nu.ght be adjusted independently they were mounted on the ends of hollow cylinders wi tt'l the slit cylinder fitting closely inside the mirror cyl;ini;ler. were held firmly by the ~dri:cal cla.n1p neA't to the slit. Both The mifror was moun-ted on ti.'le diagoneJ. end of its cylinder with .four screws so that by shimrr.ing one or more of the screws its angle could be changed. The mirror was a narrow strip of glass upon which had been evaporated aluminum. The slit was made by evaporating aluminum .onto a narrow pieee of glass upon whose surface had been stretched a 20 u. quartz fiber. After removing the fiber a 20 u slit results. To pro- tect the aluminwn coat a thin piece of cov&r gla ss wa.s fastened over it with Ce.nada Balsam. To keep the walls of the ea.mer& black and also " to cover up any holes in the aluminwn coat this cover was painted with black Egyptian Lacquer almost up to the edges of the slit. All metal parts above were Dow Metal except the mirror cylinder which we.s brass. or course various slots and holes had to be made in all these cyli.Jlders to let t.be light. beam and the barometer and fiducial arms in. The thermometer (I) was the therintl element taken from an oi"dinary oven thermometer. It was mounted on the side of the mirror cylinder. With the lever arm used the sensitivity was about 1° C per millime'ter en t:he film. The ba1•ometer was an aneroid bellows two inches in dia- metet' a.nd about mi inch long. As mentioned above i t was specially made for this laboratory by Julien P. Friez & tfons. of Baltimore. The sensitivity is about 5 - 6 millimeters def'l¢:etion per atmosphere. Since one could measure the shad0t1$ to about 2 u., the pres$ure eould be read to O. 2 or o. 5 of a millimeter of Hg. However, d'ae to other uncertainties such as variations of film devel:opment and possible .s light lags in the barometer the actual pressures are eerta.inly no better than 0.5 mm if they are that good .. ~he temperature of course could be read more accurately than necessary. • The clock situ.ation was investigated very thoroughly by both Dr.. Neher and the author. Many clocks and watches were teated f'.or fluctuations over periods of two to four minutes. Two other :f'aetors were that the timepiece had to run the eharging switch, and al;so had to: be regee.red so that its main shaft rotated o+aee in 3 J,/2 hours instead of once an hour.. Since a watch has hardly en:ou.gh power to run a charging mechanism and be-Oause both this and the regearing would be very hard to install on a watch due to its smallness and eompaetness, t a small alarm clock was :finally decided upon. The clock chos.en was , one manufactured by the New Haven Clock Company_ i4P u:ie 1a Slaell a,]Jani ele8lis.. It weighed a.bout 60 grams. The alterations for charging switch consisted merely in putting two pl.at.mum cont.a.ct post.s on a gear already iu the cloek. The gear ratio alterations involv~d practically a reconstruction o;f the mai.n. she.ft. Both of these jobs were very effici,.mtly done by Mr. S. A..Asquith, 2514 Sunse:b, Blvd., · Hollywood. The fluctuations were not more than three per cent for two minute interu-als. The drum (J) wa.s a 4 in. Dow Met.al cylinder, hollow 1.rith .the exception of a thin solid web of' the metal near one end which conneet,ed it; to .its hub. This flat, web rested upon a 111 diameter flat brass disc which was fastened on the main shg.ft of the cloek. A l)rojection f.rom the drwn fitted into arry one of 24 holes in this dise so that the ~ could be set within at lee.st 7 l/2 degrees of any position. The drum could be rerao1.red frem the sh.aft b<J merely u.n- sere,,ing a nut on the main tlhaft above and lifting it of£. The film wa~ wrapped around the 01.4tside of -this vii t.h emulsion side outward., !t wa.s held at one end by two of the very ~hort :projections of two small screws. At the other end it was held biJ two similar projections which were mounted on the end q:f a stiff piece of piano wire which ~-as fastened near the hub.. .These t wo projections ra..-i in narrow slot~ 1n the periphery of the drwn so that variations in tlle length of f,ilms, could be taken up. Th.e wire into the hub of course furnished the te.~.s ioo by wlrl.ch t.lte film. was held tightly onto the drum. The source of power (M} for the solenoid was a pair of l3ond No. 101 Plono--Cell flashlight batteries connected in serie1;.1. these batteries alone w,ould just barely work the switeh. gave a little factor of safety. One of Two in series .An investigation oi' flashli:ght 57 batteries oi' this size showed that these ,were the best.. One of the$e cells would .deliver 7 .amperes on •sliort 'c ireui t as against only 5 f:<:J~ Eveready •.t illd others. I t was found that the bat:teries dried out ~d lost their current capacity in the d.essi~ted -chamber. this they were painted with beeswax. To prevent 1'he batteries were nornw.lJ.y merely wrapped in cotton or wool iu1d wt;;1~e laid on top of the drum box. The charging SY,i tch on the clock was mE!l.~el.y ru'l. insulated strip of phosphor bronze plated with platinum (A) which norrnslly rest;ed at (B) against a spring • L][S terminal (G} of the solenoid ciri;.:uit. 'the other terminal Was the whole el.e ctro- scope. Thus when the platinum post (D) on the uninsulated clock gear CiUUe arGund and touched this strip, t11e eircuit was closed. (See Fig. 7} However, as the gear post pushed this is.trip it moved it away from its insul- ated terminal, -t.hus breaking the circuit after about one or· two eaco:ods. Eventually FIG. 7 the gear post snapped by the strip and the stage was all set for tb.e next contact about 4 minutes •later. The chamber (T) for the siccative ha.d a fine nickel screen for its top and was ordinarily filled with Magnesium Perc:hlorate. 38 Mg(C104 ) 2 • Next to P2 0 0 thi.s is the most efficient dryer known. At 25° C it keeps the pressure of water vapor down to 6 x 10-4 as against. 2.5 x 10-5 fo.r P2 o5 • • It has the advantage of having 5 to 7 times the water capaeity of P2 05 and in addition does not make a devastatj,ng liquid like Phoepheric Acid. When the condensers were behaving badly in Texas, Dr. Neher tried P2 05 but it proved no better. Dow Metal was used wherever posai ble b-eeause of its ~xtreme lightnes.a. Dow Metal F was the Catalogue designation . It has speeific gravity of 1.77 and is composed of 95.7% Mg, 4% Al, and 0,.3% Mn .. . Its melting point is. uoo° F and its co.e fficient of' thennal expansion ls 1.6 x 10...5 per degree Fahrenheit.. It is 1nanu.f'aetured by tlle Dow Chenµ.eal- Company of Midland,. Miehigan. Some parts. could not be made of this because it is pra.etieally impossible to solder. • All possible solder joints. were eliminated, but where necessary, brass v•n;:t~ used •. IV. Calibration of.' 1;..;11e Electroscopes 'l'he electroscope system used was the Neher all quartz torsion type. The construction of this ~-ystem has been written up by Millikan and Neher 24), and so will not be discussed further here. should, however, be remarked upon. Its properties Its working voltage range is 70 - 100% of its top voltage, the latter being anywhere from 550 - 700 volts. Over this range it has an al.most linear voltage C&libration. The reading .for full scale defl.eetion varies by l.ess than 0.3% between +20'°C and -20°c. It is insensitive to the directiioa of gravity and to almost all vibrations. Let it be assumed that the slit of the cam.era is exact,l.y perpendicular to the direetion of' motion of the film ~uid that the eharge sensitivity is exactly uniform over the width of the film. This is consista.nt with an aL'llost linear voltage calibration since the capacity doesn't change mu.ch. If tbe electroseope, having been filled with argon for example, is now ex.posed to some eonstan t source of radiation, the image of ~e electroscope fiber will move aeroas tl1e slit at a. uniform rate. If the film also moves at a uni.form rate, the result vdll be a sloping straight line. Since the film rate is always the same:, the slope of the line v1ill be proportional to the rate o,f motion of the electroscope image du.e to the exterior source of ionization plus tlle rate due to leak and radioactivity in the electroscope chamber itself. Since the rate of motion due to the exterior source is proportional to the rate of formation of ions in the chamber due to that source,. (4.1) km=I+Z where mis the slope, I the rate of ionization due to the external source, Z the equivalent rate of ionization due to the instrument itself, and k the constant of proportionality. of the instrument. Now k and Z are constants only If they are lmo'ku, then for aD:;f rlleasured slope m, I can be found. The constant k could be determined by measurement of the capacity of the systera, its voltage sensitivity, and the rate of mot.ion 40 of the film. Z could be determined from the asy-mtotic value which the slope approached as the electroscope is submerged in water, or else by the slope when the electroscope is plaeed in about 10 em o.f lead in a deep mine or tunnel. However, MillikBn and Cameron 25 ) had made very accurate capacity mea.aurements on several of thei.r preeisien eleetroseopes which were used fo.r under water work. They9 } had also by trial and error methods with the help of the Gold Tables found the e.ey,J1.totic vaJ.ue which the under water curve was approaching as the depth increased. With these precision instruments, accurately calibrated, Millikan proceeded to detei:mine with great ea.re the numbers of ions per cm3 per sec at 30 atme.spheres pressure in one of his e.lectroscopes when it was inside of a certain 10 cm thick spherical lead shield at a given sp0t • in his basement. This same procedure of calibration could have been followed with t;he balloon electroscopes.. It is obvious, however, that this would be a very long and di.fficult process which should be avoided if it were possible to obtain the equivalent reau.lt in another way. Moreover, it is not certain that the two electroscopes t.1-ius independently calibrated would then agree, for instance, in the lead in Millikan•-s ba sement. In fa.ct due to the much thinner walls of the balloon electroscopes one might find a slight difference between the two due to the Ubergangseffekte. This should certainly not be more than 1% for cosmic rays, bm1ever, bet-Jireen eleet.roscope walls of 0.OlS 41 in. and 0.125 in. 27 ) Therefore in order to save time and expense and also to make the balloon results exactly comparable to those of the thicker walled electroscopes, it was decided: to make the balloon electroscopes agree with Mi.llikants value of the ionization in the 10 om lead shield in his basement. This value is 25.04 ions per cm3 per sec at 30 •.l atmos- pheres of' air. To change it to one atmosphere it is necessary to divide by 15.80 which factor was determined ·by Millikan28). If ma ia the measured slope on a balloon film taken in this lead,, then (4.2) where Z now is the zero of the instrument reduced to one atmosphere. In order to determine both k and Z another relation is necessary. There is· a tunnel in the San Gabriel Canyon of about 400 ft length running under a spu.r. Dr. Neher has rambled around on top of the spur with an aneroid barometer an.d has estimated the depth -o f the middle of the tun.'1el to be 234 ft. The density of granite, of which the Sa.11. Gabriel Mountains are for-J1ed, is 2.6 gm/em3 • Using the- mass absorption law which seems t.o hold pretty well, gives 608 equivaleut feet of water or 185 equivalent meters. However, the surface above is not flat as in the case of a body of water. This makes i t somewhat easier f~r the rays to get in at inclining angles. If one approximates the gr?.nite over the tunnel by a. sphere and takes account of the uni- form 10 .meter layer of the atmosphere above, one finds for the ma,st pen$trating cosmi~ rays (u =0.05 per meter o±"" water)26) that the equivalent depth below the top of the atmosphere is 147 meters. This is certainly an absolute lower li.lnit to the depth. under water eurve one gets 0.15 volts/hr. for 185-+ 10 meters and 0.3 volttlhr. for 150 meters below the top of the atm&$phere. promise value of 0.2 volts/hr. was taken for the tunnel. A eom- Sine.a Regener•·s volts per hour are &l.'llOst exactly equal to Millikan'·s ions/cm3 sec at 50 atm. the value <>f 0.2 ions/crrl sec has peen used as the intensity of eosmic rays in the tunnel.. radioa.eti vi ty was made in the following way. The correction .t'or loeal. One of Neher" s electl"Oi:,-- s-0opes was run in the tannel, unshielded, and gave 94 ions/em3 aee • practically all of which v1as due to loc-al radia.ti@n.. Previous ~eJ:'.i- ments by both Millikan and Neher with both radium and thorium had indicated that only 0.6% of radioactive radiation registered by their eleotrosoo.pes unshielded~ was me-a.sured inside of the l.O em lead shield. This gives a ,tal.u.e of a.bout 0 .• 6 1.ons/cm3 se,c inside t-he shield due to lo~ radiation and a total .o f O..& ionSc/cm.ll .see.. The cosmic rays are o.f course unaffected: UJ' .t he lead ai.."'ter passing thro~gh so much .granite .. If ;n,_r is the slope measured on a balloon film taken in the tum.tel in lead, then ( 4.lS) C.o mbining (4.2) and (4.-3) gives 45 k : 25.04 W- O.S ) ( l3.8(mB - mr) z = 25.04 I1lT - 0.8 m;a l3.8(mB - fllT) Thus one must only determine accurately mB and lm).' in order to calibrate the electroscope against those of Millikan, Cameron and Neher. This method of calibration should give k quite accurately and Z only approxi.mt).tely for the calculations for the ionization in lead in the tunnel may have been off by as much as 20%. It is not usu.ally true that the charge sensitivity is exactly uniform and the slit exactly perpendicular to the film motion. Corrections must be made for these factors. In Fig. 8 let CD be a line parallel to the mo·tion of the film which intersects at D th.e linear electroscope trace AD. Let AC be perpendicular to CD at C and inter- sect AD at A. The measured slope of the electroscope trace will then • be m' = AC/CD. If the slit AC·• is not parallel to J~C bu.t makes an angle /, with it., the diste..nee moved by the film while the electroscope was discharging from A to D is not CD but C'D. Thus the true slope is given by If the slit were off the other way the negative sign would change to positivee The relation connecting the true and measured slopes 1i..1len 44 the slit is not perpendicular to the motion. of the film is then . (4.5) Sin-ee p will always be small, this correction will be important only w};len mt is large, but may become extremely important there. In Fig . 9 let the curve YZA be the non linear discharge curve of an electroscope e:l!..--po,sed to a constoot r adiation.. Let OX be the fiducial line and OY the dire,ction of the s lit which is taken here as perpendi<--ular to• OX. The electro$COpff then discharges iu t he negative y direction and the film moves i n the positive x direction.. If one were to mea o'Ure short sections of t his cur-ve , he would get diff erent slope.s although the ionization is the same throughout. It is obvious therefore t hat a correction f·actor must be applied to make all t I1e.se. slopes agree:. It was decided to take as the true slope of t his dis- charge, the slope measured between Y and A, t.11e t wo edges of t he f ilm. Let this be denoted by ill and any other sl.ope mea.s ured between ordine:tes Y2. end Yl as m•. Let the diff erence in ordinates between YZA and the line YA at the abs cissae :x,a and x1 be ll ;a and 6 l • m• = Ya - Yl Then m= (y~ - ~ .d - <n - .6. 1) xi - JC.a x1.;.. ~ m:;: m'(l _ A 2 - 6 1) (4.6) Ya - Y.1 It remains only to show that the ratio ~ 2 - ~ l Ya - Yt is independent of 45 the value of m.. Let the actual speed of the eleetroscope. image in they direction be Vt where t =0 wh.en y == Y, and let they speed of the hypothetical image which travels the ,atraight line YA be v . let .L1 t = the ordinate difference at ~:ny time t. Y.a = Y .... f :Also Then ;t;z. Vt dt 0 (4.7) These relations are true bees.use all points :having a.;.-:1 abscissa. Xi are e.J...-posed at the same time t. of ionization and ( v6 v) t , t-.a Because v depends only on the int6'nsi ty only on this and on the 01·dinates y1 and y2 , there is nothing in .(4.7) which depends on the speed of the film. Since by varying this speed we cen give many- desired value, it follows that ~ .a - A 1 is independent of the value of m and only dependent Ya "."" Yl on the ratio m/mt. Thus it is only necessary to determine and. plot the t, 's as a function ef y .for one value of m.. This was customarily done. for m = 1. Fig. 10 shows a typical curve., One other correction was · necessa-r-J. The intensity of cosnlie rays varies with the thickness of the atmospheric blanket. From Millikan and Cameron• .s 9 l under water cUM"e one finds that this corree- 46 ti.on amounts -to -0 .. ,15% per 1/10 in. o:f Hg f'or 10 cm lead shield at. approxin~ately sea level. Millikan. 1 a value of 25.04 ions was deter- mined for a barometric pressure of 29.26 in. Hg. All calibration sl.opes were eo,r rected to conform to this pressure. Whenever the electroscopes vtere being run in r:, place where the slope would be 1.ess. ths11 0.2, the charging switch v:as disconnected and the electroscope <".llowed to discharg-e coru1Jletely across the film . Otherwise the condensers vrnre allowed to function L11 the usual fashion . There were five instruments in all. In orde.r to l'.ttemp.t to detect possible heavy particles in the stratosphere . t wo were f'illed {Nos. 1 and 2} with l.rl.r and three (Nof:l . O, 3, 4) with argon . .It was tJ:1.ought that the.r e might be more recombination in the air electrosoopef> due to high density of ionization if heavy partieles were present .. previously The equivalence of air and a:flgon for cosmic rays h®11 been t~sted by Millikan and Neher u,p to 22,000 f't. Thi.:S i ncidentally was one of the ve:ry strong arguil'lents for the active age11t of ionization u p to t...l-d.s. al.titude be.ing electrons .. In considering the f◊llowing discussion of' the actuu CBJ.ibration, one should bear in mind that it was <lone iu a very limited ti.me, tor the ~pproximate da'te of the flights had to be set considerably in advance. Sever-al films in Millikan's basement were taken for et1.ch instrument. One f'ilm. for each was taken in the tunnel. After a run in the Sophomore Laboratory with thorium it became obvious that #4 leaked. .After it had been repaired there was no time to, do the tunnel 47 run over again so #4 was run in symu:ietey with all the others .in two places on the campus and also in Millikan' s basement. Actua.11.y, as will be shown later, No. 4 was only run against No. 0 and No. S for when subject to local radiation the air and argon electroscopes did not agree. On t..11ese last runs it was discovered that the focus of No. 5 was riot satisfactory iill1d so it was dismantled. When reassembled it was run against No. 0 and No. 4 both on the campus here and at Fort Sam Houston, Texas. After its first flight No. 4 had to be taken apart and repaired. I t was recalibrated by one fil.t11 at Fort Sam Houston and one in Wdllik&1'1 a basement after its return. The two air electroscopes ·were the ones whose condensers went bad before their flights. Therefore each of these was reduced to one atmosphere pres-s ure of' argon and used without condensers in the former manner. The only ealibratio.."l rtm possible• with these wti.s one at the standard spot a.t Fort Sam Houston because all t,he other instruments had been sent off.. As a result it wai'3 necessary to make No. l agree with the others at some point on the flights. The reason why more of the in.stru:ments could not be recalibrated after c.oming back we.s that No. 4 was the only one which was working upon its return. All of the results of these flights therefore depend upon the original calibrations of' No. 0 and No .. 3. These represent about 25 hours run in Millikan'.s basement and about 6 hours. run in the tunnel. tions is given in Table IV. The complete SU!llt"llary of the eallbra.- The following is the legend used .• - 126.6 0.89 125.7 9.84 ll.l4 ll.14 Electroscope ~ero Sophomore Lab. with Thorium On Campus near Astrophysics Sophomore Laboratory Sophomore Laboratory 17.89 $85 R.. A. M. Basement in lead with Thorium Carl Anderson's Water Wank with Thorium Above Fort Sam Houston at 4.29 meters of Water Fort Sam Houston later with pressure corrections 6.00 0.45 58.5 ,,Electroscope Constant Fort Sam Houston $/3/56 Press= 29.15 in Hg o.gsa _ o.osa : sen Gabriel Tunnel in lO cm Pb 6 .• 23 , 6.5,7 U.24 . 6,29 5.18 9.16 ll .. lS 10.97 108 0.08 7.84 .. 0.91 79.8 0.05§ 1 .. 813 1A - 0.26 62.4 l.813 l.813 46.0 4B 4A ~ U -~ § 0.25 42.6 1.815 3B 9.71 53.0 1 . 813 5A 1.813 0 R. A. M. Basement in 10 cm Pb Table IV ·45.0 . 6.29 .. 5.79 53~k lB 348 l.7.00 · 4.98 ' 8.89 7.80 107.4 0.10 74.4 2J..Q58 1.§1_~ 2A ~ 49 4A No. 4 after leak was fixed 120 lbs argon 4B No. 4 after its first flight 120 lbs argon SA No. s before being refocused 75 lbs argon 3B No. 5 after .b eing refocused 85 lbs argon 0 No .. 0 115 lb~ air 1A No. l filled with 225 lbs air 2A No. 2 filled w-ith 260 lbs a.ir lB No. 1 filled with 14.7 lbs argon Underlining denotes rtL.""lS used for determining const,ants. All values are in ions per cm'3 per sec par atm. Several factorB eoncert1ing t his table should be discussed. Fi:r·st,. 0 and 5A agree r&asonatJly w~U on all ·their runs at both high and low ioniziition. On the runs in Pasadena by which 4A was .t'itted to 0 and 3A, the consistency was quite good. In v'iew of these agreelllents which ru:·e all within 2% i t was surprising to f ind the 10% disagf<eement hetw·eeu O and 4A at Fort &'llil Houston. On tJtls run the electroscopes were set 50 ft 11ort,h of' the westhe:i:· station ou the grou:n.d. The author feels that. the radioact,ivity of the ground ma'/ not have been uniform &.i.'l.d that 4.A was more exp.osed than O. irnat.e·ver the explnnation, ·tnis disagreement weukens the accuracy of the results sornewhst because the calibra.tious cf 4B and lB depend upon the ionization at this pl.ace. Second, ll and 2A,, t he t wo air electroscopes, agreed with ea.ch other quite well on a.11 four of their simulta.neous runs bu:t showed about 20% less ionizat,ion '.f;ha.'1 the argon filled instruments. This was surpri'sing 50 in view of the fact that lfd.llikan ci.nd Neher had found that argon and air eleetroseopes which had been calibrated aga.inst each other at least partially ih lead, agreed very well not only for cosmic radiation but for , radiea~tive radiations. This discrepancy is apparently due to difference in wall thickness. The 1/8 in. walls of the large el.e ctros-copes will certainly stop all (:J particles with energies less than 2 million el.ectron volts while it should take only approximately half e million to get. through o..018 in. steel walls . . Since there are almost no natural radioa.cti ve p particles either primary or secondary of more than 2 Ulil.lion volts while there are a large number betw~en , o.5 mil.lion and 2 million, there will be many f particles which will get into the ballqon electroscopes whic·h could not enter those with t:hicker walls. T1:tis led Millikan t.o suggest that ths difference is due to the difference in recombination in &ir and argon. It is well known that .~ particles i;~ze very heavily near the end of their ranges. Many of' those getting through the walls will have lost mo.st of their energy and will '.thus be io.iJ.izing very heavily. Inasmuch a s the argon and air were at 8 imd 18 atmospheres respectively, these particles should not go far be.f ore being completely stepped. Thus most of t.h1s heavy ionization should take place near the walls where the field is weakest. Since the initial recombination in nitrogen and OX'Jgen is much worse than in argon 29 ), it is not surprising that the air electroscopes .failed t.o collect as many of' these ions as the. axgon - 51 instruments. This effect is al.so accentuated by the fact tha-t -the air pressure was double that of the argon. While this seems- to be a very reasonable explana tion, it should not be t..."\ken as fact until a special investigation has been made. The reason for comparing O and 2A subjected to thorium in Millikan' s basement i n lead imd in Carl Anderson I s water tank was to see if eliminating these ~ particles would remove the discrepancy. The results show that it is reduced bu.t not eliminated. Secondary particles from air and lead probably account for the remaining difference but again further investigation is necessary. The barometers were calibrated by enclosing the instruments in an air tight container and pumping down in stages of about 5 em of Hg t,o about 0.3 cm. The pressures were measured by a closed tube ma."lometer on the closed side of which a pressure of by a. separate pump. L 10-1 1100 was maintained Meniscus and tem11erature corrections were made so that the resultant pressure values were good to at least O.S mm. These calibrations were made at 23° C and at o° C, the latter measurements being made in the Cold Room of the Biology Depart,ment. In most cases this ma.de a difference in the total expansion of the bellows which corresponded to about 6 mm of Hg . No runs were ma.de at hi gher temperatures because it was expected that the temperat ures would be low in spite of attempted compensations .. A brief discussion of the actual temperatures encountered and their possible effect on the pressure readings will be given below. 52 The thermometers were ca.librated °b'J enclosing the electroscope in a Balsa-wood box and cooling v:ith solid 00.,a. The temperature was lowered to between -10° and -20° Centigriide in a.bout 3 hours. A thermocouple with one end connected to the inside of the electroscope next to the thermometer element., and the other to a potentiometer gave the true temperature.. The light was ·turned off at each reading in order to mark the f"ilm. V. Flight Technique .An electroscope which was to -be s.ent off was prepared for its flight in the following way. Fresh siceative was placed in the drying chamber and :f'resh dry batteries connected to the solenoid. The clock was suit&bly wound., the film loaded in the usu.a.l manner, and the hemi~~ spherical cover fastened on with screws. In most ca.se.s in order to keep the inside dry, this joint as well as all screws which connected with the recording cru,..mber were painted with beeswax. After the con- denser had been charged and checked through the eyepiece opening, the cap (U} was screwed on loosely in order to provide air passage for the barOllleter. A small reflector was .fastened to the windo,, {A) at such an angle that when the ion chamber was above the recorder, the light which exposed the film came from slight,ly below the horizontal,. This insured light from the earth and sky near the horizon .. This light should be quite constant both in intensity and in speetral di.stribu.tion, whereas skylight from overhead gets less and leS$ strong as 53 the top of the atmosphere ls approached. Part of the heat insulation was next provided by wrapping the whole instrw-nent with about l in. thick white wool. In order to pro- tect the instrument in case of a hard landing; it was supported in a basket made of l/16" steel wire, thoroughly braced with finer wire. The electroscope was supported by six steel wirea of the same kind. These were fastened by screws to the junction of the t wo electroscope • spheres and also hooked on to the basket, where they were wired in place. 'l'o provide for the remainder of the heat insulation the cage was completely o-overed with cellophane. Thus besides the ,voolt a layer of :fairly still air surrounded the instrument. One quarter to one third of the cellophane was painted black so that it would absorb some of tile ~'ID."1.' s radiation filld. sµ.ppl y.: a. little extra heat. This was merely an estimate since the exact calculation of the problem of' temperature control is Virtually impossible. be discussed later. The extent o~ success obtained will The electroscope was then •ready for its flight and was kept upside down until the serid-off in order to keep the clock stopped. The balloons were 2 and 2. 5 meters in diameter. e.nd were .com- . posed of sectors of very thin rubber cemented together. They were pro- V'ided with a thick rubber pad 1 foot in die.meter surrounding the appendix. There were in addition five lugs on the balloons, one o-n top and four evenly spaced around the equator. It was natural to suppose that the load was to be attnched to the four equatorial lugs. 54 and that the top lug was merely for support while inflating. This was the method used in all the 1952 flights by both this laboratory and by Regener. When this time the balloons did not seem to give proper altitudes, the load was suspended in one case from the appendix and in another from the top lug. Later it was found that the appendix actually is the proper point of suspension. In two eases t wo 2 meter balloons were used with the loads suspended from the side lugs. The balloons were made by the Continental Caoutchouc Company of Hanover, The weight of the 2.5 meter ba.11.oons was about 10 lbs and Germany. the 2 meter somewhat less. A six foot square, bright red, China silk parachute was provided to bring the instruments down when and if the balloon burst. The red color was · ehosen to attract attention and so to make the recovery quicker and sur~l?. The top of this parachute was provided with a release tci"' clear it from the remnants of the balloon. Four strings extended from the corners of the parachu,te to a 1 ft wicker ring which acted as a spreader. Below this was usually hung the eleetroscope~ although. a vane for turning the instrument w-a.s instlrtcd 1·0:..· ..n. ... flight bat later found unnecessary. As soon as the morning fog cleared the inflation was begun. The balloon was filled with hydrogen until it would lift weight equivalent to the actual load plus 1200 g.!lls free lift, if it was a 2.5 met.er · balloon. In the case of the 2 meter balloons 800 grams free lift was provided. The resultant rate of rise was 600-700 ft per minute. When 55 the inflation had been completed, the parachute and instruments were fastened below and the balloon released. The man holding the instru- ments had to :\{.eep neerly under the balloon until it was high enough to swing the instruments cl.ear of the ground. In these flights very little wind was encountered. Almost the last act of the instrwnent carrier was of course to give the electro·scope a sharp t wist to start the eloek. Three theodolites were used to follow the balloon. and measure altitudes. With suitabJ.e. Arrrzy- publicity t..rirough<;>ut the countryside, it was hoped tha.t recovery would be speedy. Fig. ll shows the inflation of a 2 meter balloon. Fig . 12 shows the electro- scope and para.chute ready for the ascent. Fig. 15 shows t wo balloons with para.chute and instruments shortly after the send off. VI. Results. g! ~ Flight§! Six flights were made in all. Electroscope No. 4 was found on the same day it vJa.s sent out and therefore was recalibrated and sent out again. Table V gives a SW!hllal7 of the results of the flights except the actual ionization measurements. Of tho tr.ree 1)e.ll0ons which did not come to rest with tne instr..1:ments, t wo were found. Since a.11 six instrument.a were returned eight out of' nine possible recoveries were made or 88 per cent. One outstanding fact revealed here is the very poor altitude reached in all the fiight a exc~pt the one on . August 7, 1955. It is believed ·that the rubber deteriorated in the long trip from Gerr!lany and in the heat encountered in T~a.s .. Tao,ie V ' . • . :.~ • - Elect:ro.ec>pe #0 , . i'.; -. ·, 2 mete:l .BallQon , Elec:rtrc,$eope: #5 • . , 2 1/2 meter &!loon,, . .: ' Electros~c,pe #~ , '• ·1 2 l I 2 fl\e;lie.r . i#l ·.~ Electroscope-#~:,,,\ , 2 l/2 m~ter• ·~99n· .I. • ; •... :'~~-> // ~-: ~ Electroscope ill. . • '• \ 2""'2 meter •a.~"p. ~n,'·. • . .. ~-;:•··· .. . •• .·:··-r·;r~_.: ' Eleetrosoope'. '12: 2•2 meter Ba~QQJ\¥~, • Radio Trattsnti ttet, +:I ~ (I) Q) tb f/.r.i ii f !1 At ti) / (D ~ ~ +» Q) 0 ID t:t 'gFt ·(!) e.. Q3 ,.<;.j ..... ~ (l) a, ........ Q) .....a ~ 1'0.! Ii!'; m J:: ~ :s ~ :i ~ Q) t1 i <U B e ~ ~ r'◄ Cl) ; '" ~ Q) ~ () © Q) l\lix; .~ g- '"4 0 © ";f 00 i () tb Q) +) ~ Ci! ' !!,IM 11:-t 'Q ~ Q) a ! .t 52°0 ... (!) 29°c +)© ,:i .... umul::eau~ht ◄ O 302 0-29,000 .' ' ' ' ~1,~:~~ij.~., ;) .· ssoc 18°c. Atiet4e~,e;,Jt ... 54oc 5-foc 15°c aU;,;~~;~ ~g¢ E\,~,~ ~~tl~i>e 530c 00,ndeuser . :: ,, ... · .....·'·.' .. • , ' e, ' e 0-58,000 ' 2.5 501 501 ' ; _ ' .' ( 45 43 525 553 85 50 B5°C B0°c ~ 112 325 None 50 • 47 Q) ~ ·! c:, a1°c 15•c 11 t,? 36°0 - Cl) H l>:l Oondenoer ... ! ~ 0-59,000 Not exoosed .. . ...... I... is :&!!~ ·aj 358 546 Noae 1 g 1 I Ii 102 510 510 46'~000? . ... :fj! .' b ' 0 A - 25 25 36,200 59.000 } 42.,000 Cl) A 45,000 56,000 +) 8-4 10-18 8-4 - 502 (9-l) ~ 8-6 11-17 . ll-17 . 83 (l) +) 9... 7 19,00029,000 8-12 (J:0..10) (l0-10) .. 6$ ,. ·000 · ·,, 8-20 8-26 41.200 ' - : . 8-7· 8-7 8-22 . 9-4 (9-8) 56 It is significant that only t wo out of eigl1t balloons burst and even those at the unconventi.o nal altitudes of 36,000 ft and 21,000 ft. In all other cases the balloons merely sprang leaks and descended slowly. It is interesting to note that in three eases the balloons und instruments were found together, while in three more they were not. On the flight of August 7 and August 20 the balloons brougbt the instruments down and collapsed with the11. • On the flight of August 6 the balloon burst at 56 1000 ft and being tied too close to the electroscope fell on it before the release could work. In the crash the electroscope system was broken bu..t the clock continued to run and the .f'ilm was undamaged although the bottom of the electroscope was badly smashed in. This speaks extremely well for the protection given the film by electroscopes of thi s design. In the flight of .August 12 release the parachute,/lpparently let go a.t about 10,000 ft on the descent"" This electroscope came back in perfect condition with hardly a hole in the cellophane, thus showing the size of parachute to be sufficient. The balloon has not yet been found. In t..li.e two other eases where the bal.10011 came to rest far from the instruments, it is quite apparent that the instruments in striking the ground before the balloon, w~re disconnected when the para.chute release worked. The balloon thus reliev~d of its load rose r apidly and traveled 50 - 70 miles further. That a.ll the flighta except one landed in the northwest quadrant was surprising at thi3 time. The meteorologists. however, say that such a phenomenon is to be expe,e ted in the sum,11er in Texae. In 57 no flight was the ionization measurement successful to the matimum altitude reached. The condensers caused most of the difficulty. However, on the one really high flight (August 7), the condensers w-ere Yrorldng well . b11t the drum eaught, stopping the clock at 50,:000 ft. The satisfactory performance of the condensers cm this flight was pro'bably due to the temperature being the lowest of any flight. This in tum was due to the fact that the send off was 9:25 A.M. instead of the usual 11 A.ii. to 1 P.M. In general the instrument temperature rose a little immediately a.fter release, in every case reaching a max.... imum 1° to 4o C above the starting temperature after about 20 minut.e s of flight. The temperature then fell steadily to a minimum when the el.ectroscop.e was again nearing the ground. about 20° C per hour at high levels. This rate of decrease was For a higher flight this might take the temperature too low. Therefore the cellophane will be painted half or more black on the next trial. The curve obtained. ln these flights of ioni:::;a.tio11 against depth· below t.he top of the atmosphere is given in Fig. 14. Each of the points plotted represents a 4 minute reading ex.e ept some of two minutes or less near the top. All the eorreetions mentioned above have been made. There .remain, however, several sources of error. Their magnitude and the means taken to correct for them should be mentioned. Syetem&tic Errors... 1.. There is some possibility of a mechanical -lag in the a."leroid bellows,. Tha,t this is small has been sho¼u in calibrations 58 both by Julien P. Friez and by Dr. Neher. The latter finds when one of these aneroids has been taken down to one third of an atmosphere and then returned to atmospheric pressure, that a half hour after the return the barometer register-a within l millimeter of its reading before the pressure red:iction. By decreasing the pressure during cal- ibration at approximately the same rate as it decreased during the actual flights, this error was furt..11.er eliminated~ It is probably negligible compared to other er1·ors l:llld fluctuations. 2. There may have been a slight barometer lag due to the failure of the pressure fall inside the recording chamber to keep pace with the fall outside. If this difficulty existed it was because the recording chamber was sealed too tightly in order to keep its humidity down. It is impossible to estimate this error Cbr the the-0dolite heights were not dependable enough. With the humidity problem solved, it will be easy to eliminate t,his ~cer'.:ainty in fature .flights. 5. A temperature correct,ion should be applied :to the barometer .. It was not possible to ealeulate this because no calibration rl:ll'ls were made at high ter..1peratures. Judging from the temperature correction between 22° C end o° C this correction should certainly be lea$ than 0.04 meters of water in every case .and probably negligible ill the two flights of No. 4. l..n idea of the magnitude of this correction will be gained upon rec-aJ.ibra.tiQn. 4. In spite of the carefully planned method of ca.llbra·tion, there is still a chance of eJTQr due to vrall eff'ect .. Although these 59 thin walled instrwnents were made to agree v;ith the thick walled ones for the hard cosmic rays inside of 10 cm of Pb, they might not neccessarily agree for the sort cosmic rays aloft. Since these instruments a.re for use vd.th these softer rays it might be advisable to calibrate with softer rays. Both methods will be tried before the next flig.,.11ts to see if they agree. In this respect these thin walled chambers also differ from those of the Dallas flights of Bowen and Milllkan1 8) which were calibrated with ¥ rays fro~ radioactive substances. 5. • A systematic bw:ometric error may have been caused by a difference in the shrinkage of the calibration films and the flight films caused by the difference in their histories before and during Correction ha.s not been made fof this. development. would ·be to make the pressure readings too high. Its tendency In the future an attempt will be made to take aeoO'Wlt of this as well as the differences in temperature and wetness of the f4lm when it is measured. 6. No temperature correct.ion for the recombination in argon has been applied because it has been found that for the pressures used here and the small temperature variations it is negligibly small. Random Errors _ 1.. The lack of sufficient calibration in Pasadena. and the unex:peeted disagreements in the value of our standard ionization at Fort Sam Houston allow an uncertainty er several per cent in the constants f:fJ all the instruments except No. O. Thismey have ca.used a 60 random error of a few per cent. 2. Clock nuctuatioos probably account for a large pa.rt of the scatter of the points of' individual instruments. Some of the clocks were tested and found to have a mean error of about 1% for a two minute period and a maximum of 5 - 4%. For a four minute run these would be somewhat le,s s. Convection currents cause the electroscope f'ibre to weave, 5. thus .. making wiggly lines and adding cousiderable uncertainty to the slope. Fortunately, the method of heat insulation used eliminated con- vection al..rnost completely ~cept ta:C the v-ery beginning of the flight. In view of these various sources of error, the values of the ionization represented by the curve may be in error by as much a~ 10 per cent. The shape o.f the curve is probably somewhat bett.er (5%) though the fact that some of the electroscopes only contribute points over short portions of the curve ma;z t en<l to distort it a little. In vie.,1 of the contention made above~ that single discharge el.ectroseopes are not as accurate as those using multiple discharges., it is fitting to discuss briefly the spread of the readings of electroscope No. l without condensers as compared to the spread of the other instruments. At pressures grc:ater than 4.5 meters t here is 110 question but that the internal consistency of No. l is much worse than that of the ot,hers. At lower pressures if is as good or slightly better. surprising result is due to three different causes. This First. the film from No. l had by far the most contrast of any film, thus ma.'ldng its 61 measurement easier and more accurate. Seeond, the top few values on No. O and No. 4, s-econd flight,, were for lengths of time much less than 4 minutes because of the low voltE);ge of the rapidly dischargj.ng condensers. Considerably better consistene:o.r would have been gained with longer discharge because the clock f'luctuat.ions f'or four minutes are much less than those for 2 minutes. Finally, and most important, No. 1 went off scale at but 10,000 .rt after only l hour and 10 minutes 0 of flight. Since it is hoped with better balloons to reach an altitude of 80-90,000 ft in a. 3 hour flight, the sensitivity of No. l would be f ar too high. It should be decreased by a .fnati<i>f of 4 or 5 at least. With the sensitivity thus decreased, the uncertainties which occurred at pressures higher than 4.5 meters on t..1-iis flight would run all the way to 2 meters. ·Thus these flights bear out to some extent the need for a higher sensitivity than single charges will allow. In Fig .. 15 are plotted besides t-1:>e Fort Sam Houston results (FSH), other very high altitude results obtained previously by Bowen, M.i lllkan, and l{eher of this laborator.1. The curve MF up to the c:ros.s represents the result of aeropla..vie flights from March Field ( 4IN). The curve P represents sounding balloon fli ghts from Dalla s (421'1') and the curve (SF) is that of t..11e Settle Fo.rdney balloon flight (52N). At the time the Dallas results were publi shed, very small accuracy was claimed for them. The l!"SH curve indicates that up to 2 meters these re~t-s were essentially correct. However, the FSH curve at 59° N should fall slightly below both the 41N and 42N curves. Instead it is 62 5 to 4% above. The discrepancy with the Dallas curve is not of' great . importance since it is for less than the combined experimental errors. Neither should t,he diserepanoy with the aeroplane curve be ~oo disquieting in vimv of the possible sources of error mentioned above. These electroscope flights did not produee any startling results as f'ar as the study of cosmic rays is concerned. They agree well enough.with the previous results11) that the Gold analyses applied there need not be· done over. Since the previous flights a method of analysis has been developed by Gross32} and its experimental verification is claimed by Regener and Pfotzeri5). This analysis has been used by Gross 52), Regener and Pfotzer55 ), and Compton16) 34), to bring out humps in some of the ionization altitude curves.. These humps are attributed to magnetic range minima of electrons and o< particles entering the eerthts atmosphere from without. Whether this analysis is valid or not, the author does not believe, in view of the errors mentioned above, t.hat the shape of the curve obtained by applying such an analysis to the Fort Sam Houston data can have any significance. The results of these flights therefore say very little about the validity of these humps. Within about 5%, however, there do n.ot seem. ~o be any irregularities in the curve. In conclusion, the author Vliehes to thank the members of the department and shop and the graduate $tudents, all of whom have been most helpful in many ways. In particular, the sound advice of Dr. Bowen, the able direct.ion of Dr. Millikan, the excellent inst.rument 63 design and cons-truction of Mr. Pearson are de.e ply appreciated. The author was exteemely fortunate in and is most thankful for being able. to work with and. immediately under the direction of' Dro Neher, without . whose generous help very ll t·tle progress would have been m.ad,e. .ill of this work was made possible through a grant from the Carnsgie Corporation of New York administered by the Carnegie Institution of Washington. VII. l. Summar-,r For use as a source of electroscope pote."ltial., condensers having leakages of less than 1% an hour at room temperature were developed,.. Later because 0£ the failure or these at high temperatures a new type of condenser was developed, haV'ing a leakage l.ess than 1% per hour for all temperatures from 2° C to 47° o. 2. The coefficient of thermal expansion of ceresin was roughly measured betv,een o° C and 20° C and found to be 2.5 x 10-4 °c-l. A large dis- crepancy bet.ween the tabular values for the resistivity of mica .and the values obtained from the leakage of eondense:rs was found. This invites further investigation. 5. The desig-n of' the Neher automatically recording electroscope has been modified to make possible its use f or soWlding balloon flights. 4. Five of these instruments were constructed, calibrated, and flO'.r-m from Fort Sam Houston, Texas, Magnetic Latit:u!le 59° If,. using the earlier type of eon.denser. 5. Because of the failure of condense-rs and balloons, ionization records were obtained only to 40,000 ft. 6. These results check those of Milliken and Bowen from Dallas in 1952 vd.th regard to the shape of the ionization depth curve. A dis- crepancy of about 5%in absolute values both with the De.lla.s results and the aeroplane values from Marci1 Field indicates the.t method of ce.libration ahou.ld be further investigated .for effects of wall thickness. 7. Within a limit of about 5~ whicl is set by the scatter of the points, there are no irregularities in the curve between 7 meters of . . water and 2 meters of water. ·. At depths greater than 7 meters of water the data. are very inaecura~.. a. The advantage. of repeated charging has been demonstrated. 9. Possible systematic and random errors in the ionization and pressure values are discussed with suggestions for future flights. lO. The following features were found to be somewhat unsatisfactory and will be or have been altered. a. The balloons did not rise high enough because of leaks. b. As mentioned above, the condensers failed. c. The super-sensitive panchromatic film .fogged too much in the heat. Film which is not red sensitive will be used • hereafter. d. Tile temperature compensation was not quite sufficient. More of the eellophane will henceforth be painted blaek. 65 References l) 2) J. Clay, Proc. Acad. Sci. Amsterdam, 50, lll5, 1927 J. Clay, n J. Clay, " tt ff " 51, 1091, 1928 ff $3, 711, 1950 A. H. C.o mpton, Physical Review, 41, lll, 1932 41 , 681, 1952 45, 387, 1935 5}' H. Hoerlin, Nature, l.52! 61, l.953 4) R. A. Millikan, Physical Review, 43., 661, 1955 R. A. Millikan and H. V. Neher, Physical. Review, 47, 205, 1955 5) J. Prins, Nature, 152, 781., 1955 6} J. Clay, Physiea, l, 565, 1954 l, 829, 195L1 7) W. Kolhorster, Berlin Berichta, 54, 566, 1925 8) • E. Rege;ier, Die Natu.rwessenschaften., 15, 185, 1929 E. Regener, Zeits. fu Physik, 74, 435, 19:55 9) R,. A. Millikan and G. H. Cameron, Physical ReView, 57, 235, 1951 10) J. Clay, Physica, 2, 645, 1935 U) I. s. Bowen, R. A. Millikan, and H. V. Neher, Physical Review,. 44, 24:S, 195~ 12) I. s. Bowen, R. A. I\tlllikan end H. V. Neher, Physical Review, 46, 641, 1954 15) Unpublished results o.f Ndllikan and Ueher in Peru and the Philippines 14) L. M. Mott-Smith and L. G. Howell, Physical Review, 44, 4, 1955 66 15) A. Piccard and M,. CoS'JnS, Comptes rendus, 195, 605, 1932 16) A. H. Compton and R. J. Stephensen, Physical Review,. i:15, 441, 1954 17) E. Regener, Physik Zeitschrift, 34, 5G6, -19-05 Naturwissensohaften, 20, 695, 195.2 . 18) I. S. Bowen and R. A. Millikan, Physical Review, ~!, 69f>,. 1955 19) I. s. Bowen and R. A. Milliken, P:b:ysicru. ReView, 27• 555, 1926 20) J. Clay, Physica, l, 365, 1954 21) Curtis, Bulletin of the Bureau of Standards, ll, 559, 1915 22) John Strong, Astropbyei.o al Journal, June 1936 25) U. Retzow, Die Eigensehaften elektrotechniseher Isoliermateria.llen in graphischen Da:rstell:ungen, Page 81, Fig. l:?>6 24) R. A. Millikan and H. V.. Neher, Probably will appear in th-e Physical Review about June 15, 1956 25) R. A,. Millikan and G. H. Cameron, Physical Review, 51, 921, 1928 26) 27) Heinz Schindler, Zeitschrift fur Physik, 72, 625, 1931 26) R•. A. Millikan,. Physical Review, 39, 597, 1952 29) E. F • . Cox, Physical Review, 45, 503, 19~ 50) Lewis Wright, Light (Macmillan Co., 1892) page 28.9 51) J~ Strong, Rev. Sci. Inst., 6, 243; 1955 32) B. Gross, Zeitschrift fur Physik, 83, 214, 1935 55) E. Regener and G. Pf'otzer, Nature, 156, 718, lS.3$ 34) A. H. Compton, Rev. Sei. Inst.,. 7, ?l, 1936 a.. WIDE b. NARROW PLATE PLATE A ,-------,/ , ·---~- - ---- ___ ,.__ __________ _ I I .........._......._.~ c. FINI SHED CONDENSER Fl G. I