326 MONTHLY WEATHER REVIEW SEPTEMBER 1937
under a pressure of 40 cm mercury and to 70 percent absorption under a pressure of 9 cm. This behavior is quite different from that of water vapor. It can be in- ferred from theoretical arguments that the individual lines which compose the absorption band of COa must be
rather close together. One can prove that if the cores of
the lines do completely overlap, the absorption becomes independent of pressura. In the COz band this condition
is approximately realized. For computations of atnios-
pheric transmission we may neglect the pressure depend-
ence of the absorption in cases in which no great precision
is required. There seem to be no other constituent,s of
the atmosphere which absorb, appreciably, infrared radia- tion; ozone (19) in the concentrations in which it is present
in the atmosphere is almost transparent in the far infrared.
CONCLUGIONS
In the preceding section we have tried to give an exhaustive enumeration of all the effects which under
atmospheric conditions could possibly modify the absorp-
tion coefficient of water vapor. The main effect is the dependence upon pressure and temperature treated in the
first section. There are two accessory features to be taken
into account. The first is the appearance of gaps of partial
or complete transparency in spectral regions which in Simpson’s method are still treated as absorbing continu-
ously; this refers especially to the region beyond 1511.
The gaps will appear a t places intermediate between two lines; these effects can be calculated from the formulae
iven above, when somewhat more precise values of the h e intensities are known. The second effect is the change
of the total line intensities due to the change in concentra-
tion of the corresponding molecular states. All the effects which have been mentioned in this
paper tend to decrease the absorption coefficient as com-
ared with the values used so far, especially in the higher
f evels of the atmosphere. Our results lend themselves to a
number of applications concerning the absorption of ter-
restrial and solar radiation by the atmosphere which will be given a t a later time. We might c.onfine ourselves here to a few prelimina.ry remarks. Simpson assumes in
his work a content of about 0.3 mm of precipitable water for the whole of the st,ratosphere. Now under the reduced pressure this layer has an absorpt,ion corresponding to
less than one-fift,h of this amount of water under norma.1 pressure. It follows that the stra.tosphere is practically
transparent with the exception of the very narrow spectral
regions occupied by the line cores. The flux of radiation through the stratosphere takes place principally in the
cores; the radiation is therefore of a different spectral composition from that emitted by the troposphere. No direct inferences about the thermal state of the stratosphere
can thus be drawn from a knowledge of the radiative
transfer in the troposphere.
REFERENCES
(1) H. R.ubens and G. Hether, F.’erhand. deitt. phys. Ges. I S , 149 (1916); G. Hettner, Ann. der. Physik, 65 476 (1918). (2) F. E. Fowle, Smithsonian Ilfisc. Coll. vol. 68, d o . 8 (1917).
(3) L. R. Weber and H. hi. Randall, Ph.ys. Rev. 40, 835 (1932).
(4) G. C. Simpson, Mcm. Roy. Meteor. SOC. vol. 3, No. 21 (1925). (5) G. C. Sinipsori Mcm. Roy. Meteor. Soe. vol. 3, No. 23 (1929). (6) C. G. Abbot, dmithsonian Misc. Coll. vol. 83, No. 3 (1929). (5) H. Wesler, MONTHLY WEATHER REVIEW, 64, 122 (1936).
(8) R. Mecke and others, Zcits. f. Physik, 61, 313, 445 and 465
(9) See e. g. Leigh Page, Introduction to Theoretical Physics,
(10) See e, g., A. Mitchell and M. Zemansky, Reson.anceRadiat,ion
(11) See V. Weisskopf, Physikol. Zeits. Sd, 1 (1933).
(12) D. M. Dennison, Phys. Reo. 31, 503 (1928). (13) See e. g., Geiger-Scheel’s Handbuch der Physik, vol. I X , p. 399. (14) E. I<. Plyler and W. W. Slestor, Ph.ys. Rev. 57, 1493 (1931). (15) H. Bussmann, Zeits. I. Phys. 4S1 831 (1925). (16) H. Becker, Zeits. f. Phys. 69, 583 (1929). (17) H. Rubens and F. Ladenburg, T’erhandl. deut. phys. Ges. 7 ,
(18) G. Hertz, Verh. dcut. phys. Ges. I S , 617 (1911). (19) G. Hettner, R. Polilniann and H. J. Schuhmachez, Zeits.
(1933).
Ch. XIk.
and excited Atoms.
170 (1905).
f. Phys. 91, 371 (1934).
ON PILOT BALLOONS AND SOURCES OF LIGHT FOR HIGH ALTITUDE UPPER-WIND
OBSERVATIONS
By WILLIAM H. WENSTROM, Major, United States Army (retired)
[New Haven, Conn , October 18373
The research described herein was begun in December
1931, in response to military meteorological problems, and
continued as occasional opportunity offered until Sep- tember 1937 when the work was necessarily terminated on account of administrative reasons. Due to these limitit-
tions, some of the data obtained are incomplete, and the
results presented should be regarded as first approximn-
value to meteorologists, because the field has previously
been little explored. When cloudiness or poor visibility exists in the lower
levels, upper wind determinations to high altitudes can be made only by means of the yet experimental radio pilot balloon,’ or by means of expensive military techniques.
The radio pilot-balloon problem is a complicated one, and its two-dimensional solution with a desirable precision of
direction (X o of 1 degree). will probably require many
years. The research described herein is concerned only with the far more simple problem of deternlinirlg upper
winds during the day or night to altitudes of 15,000 to
1 W. H. Wenstrom, Radiometeorography BS Applied to Unmanned Balloons, MONTHLY WEATHEB REVIEW, vol. 62, July 1934.
40,000 feet in good visibility under clear skies or high cirriform clouds. Recent aviation trends toward flying
in the substratosphere, 20,000 to 30,000 feet, have brought this problem into immediate practicar importance. In
addition, high-altitude upper-mind observations are often
useful to a weather analyst and forecaster.
tions. They are nevertheless thought to be of general PILOT BALLOONS
S t n d u r d 6-inch pilot balloon.-The standard 6-inch pilot balloon, I ~S used by the United States Weather Bureau
and the Army Meteorological Service, weighs about
1 02. (30 9) and costs about 10 cents. Inflated wit11 hydrogen (cost &out 10 cents) to a free lift of 4.66 OZ.
(137 g), corresPondi11g to a sea-level diameter of about 2%
feet, it rises a t an average rate (except in the zone of lower-level turbulence) of about 200 yards per minute.
IThite, red, and black colors are available for use, respec- tively, under clear sky, cirriform clouds or darker clouds.
A normally inflated balloon usually bursts a t an altitude around 30,000 to 40,000 feet; but due to its small size,
slow rise rate, increasing winds a t upper levels, and the
SEPTBYBBR 1937 MONTHLY WEATHER REVIEW 327
optical limitations of standard theodolites, even under clear skies the balloon is usually lost to view (unless the
upper winds are exceptionally light) a t an altitude be,low .~
20,-000 feet. Balloon clusters and day beucons.-During the winter
of 1934-35 a t Bolling. Field. D. C.. and Aberdeen Proving
Ground, Md., ballock clusters of various sorts were tried
in an attempt to attain higher altitudes than were prac- ticable with a single 6-inch balloon. A cluster of three
or four white or red, or both, balloons, snugly tied neck
to neck, was tried; also a tandem cluster of three balloons spaced about 20 feet apart along a string, the upper
balloon being somewhat overinflated ; results were slightly better than with a single balloon. Somewhat better
results were obtained by hanging a “day beacon” (a large open cylinder of glazed white paper on a frame of No. 16
copperclad steel wire, size about 20 by 30 inches, weight about 2 oz., reflecting sunlight better than the balloons) about 50 feet below the cluster with a string. It is
possible2 to connect some device such as a water-filled
copper tube (or perhaps a sealed glass rial) in the sus- pended string, so that the day beacon will drop when the
ireezing (0’ C) level is reached. At night we used some four-candle lanterns (fully described below) with 6-inch
balloon clusters. AI1 these e,arly experiments served to show, however, that what we really nee,ded was a larger pilot balloon. A
9-inch dot balloon was used for high-altitude work by the
Army beteorological Servic.e during the war, but wgs later discontinued. Large sounding balloons were avail-
able, of course, . but were prohibitively expensive for ordinary upper rmnd observations.
Twelue-inch pilot balloon.-During the winter of 1934-35
Maj. A. H. Thiessen, Chief of the Army Meteorological Service at that time, set up specifications for a 12-mch pilot balloon, to be manufactured b a process of coagula-
characteristics:
tion of latex on outsidc of spherica 9 mold. Approximate
Twelve-inch pilot balloon (day) :
Balloon weight: 29i 02. (80 9).
Free lift: 151i oa. (430 g). Average rise rate: 300 yd./min. Average bursting altitude: 40,000 to 50,000 feet.
Four-candle lantern weight: 5% 02. (150 g). Free lift: 18 02. (510 9). Average rise rate: 220 yd./min. Average bursting altitude: above 25,000 ft. Also used with 7% oa. 8-ca.ndle lantern, rising at 225
During 1935 and 1936 about 250 ascents were made with this balloon a t Bolling Field, Aberde,en Proving Ground, and a.t the temporary Stratocamp in South
D a k ~t a .~ Best results were naturally obtained in the clear air of the West. The 12-inch balloon c.ould be seen
farther away than the 6-inch balloon, and its gre.ater rise rete resulted in. higher altitude for a given distance out.
Double-theodolit,e day runs to 40,000 feet were easily
made in light upper winds; heights greater than 20,000 feet could be reached even in high winds. At night the
1 %inch balloon with a 4-candle lantern performed. con- siderably better thnn the 6-hc.h balloon with a single-
candle lantern; one run under favorable conditions was
followed (by single theodolite) to an (assumed) altitude
of a.bout 27,000 feet above station or 31,500 feet above sea level. Large pilot balloons of this type, made by the latex -coagulation - on -outside - of - sphencal - mold process
though expensive (perhaps $2 per unit; hydrogen cost
(Night) :
yd./min. when inflated to 22 0 2 .
9 J. F. Brennm-“A method of determining the altitude of the freezing point”-Mo. Wen. Rev., vol. 6% February 1931.
perhaps 30 cents per unit), are uniform in size, shape,
end thkkness. Their rise rates are correspondingly uniform, and suited to single-theodolite observations.
Sil:teen.-i.nch. pilot ba,lloon.-About 1936 a 16-inch (100 g)
pilot’ balloon becanie amilnble. The process consisted of
spinning htex t o general bnlloon shape inside a small spherical mold and expanding the hollow ball of latex
thus fornied, while still soft, to full 16-inch bnlloon size
by a.ir pressure. Compared with the older solicl-spherical-
mold process, this air-expanded process produc,es ba.lloons
that are considerably lighter and far chea,per (perhaps
50 cents per unit; hydrogen cost perhaps 50 cenhs per
unit), but apparently less uniform. The balloons tested,
a.t least, were not very uniform in size, shape or thickness;
later models, it is understood, have been iiiiproved con-
siderably in these respects. PrwAkally, such ineqdities result in an appreciable percentage of. premature bursts;
and (despite uniform inflation) in c.ons1derable and unpre- dictable variations of rise rate, necessitating double-
theodolite observat,ions. Far outweighing these disad-
vantages, however, is the high rise rate and lifting power
of the 16-inch balloon.
During 1936 and 1937 we made about 200 test ascents a t Rollirig Field and Aberdeen Proving Ground with the
16-inch balloon, asc.ending bot>li by itself and in combina- tion with various weights and light sources. Preliminary
tests showed that rise rates around 400 yd./nlin., attained
with free lifts around 40 oz., were optinium; lower lifts
and rates did not materially reduce premature bursts, while higher lifts and rates clid increase them. As 400
yd./rnin. is a round number exactly twice the standard
rise rate for 6-inc.h bnlloons, it was soon adopted as
standard.
The highest free lift used was about 44 oz., which in
winter gave rise rates around 400 yd./inin. a t the lower levels for the balloon alone. Above 10,000 feet this rise
rate usually inc.reasgd gradually, perhaps by 10 percent
a t altitudes around 30,000 feet, the increase being ap-
parently due to some decrease in tumbling and oscillation as well ns increase of the volunie/a.rea r h o . I n summer
the rise rnte wa.8 higher, averaging 450 yd./imn. in the
lower levels for a free lift, of 42 oz. ; the upper level inc.rea.se
was less marked, perlinps due t o steeper lapse rates in the summer a.tinosphere. Differences bet#n-een the average
rise rate of individual a.scents amounted to 10 percent or so, occasiona.lly t80 20 percent. During an ascent, the rise
rate was likely t o vary from level to level by 10 to 20
percent. As a working free lift t o approshnate the 400
vd./Inin. rise rate. through all levels under most conditions
\vith the 16-inch bdloon alone, 40 oz. a.ppears tp be the best figure. Summaiizing nppraimat,e c.hn.racteristics:
16-inch pilot bdloon (day):
Balloon weight: 3 3 08. (100 g)
Free lift: 40 02. (1,130 g)
Average rise rate (balloon only) : Around 400 yd./min. Average bursting altitude: 40,000 to 50,000 ft.
dbout 100 daytime ascents were mnde with the balloon
a.lone under va.rious conditions, a t least 30 being checked
by double t,lieodolite t.0 altihudes above 30,000 feet. Under clear skies and through average upper winds the
wl&e bnlloon could be followed with stnndard theodolites to distances of 20 or 30 miles and to altitucles of 40,000 to 50,000 leet. One observation in March extended to
36,000 feet altitude through winds of 90 ni. p. 11. at 10,000
ft., 70 m. p. h. at 20,000 ft., and 120 In. p. h. a t 30,000
3 W. H, Wenstrom-“Some Interesting Pilot Balloon @bsersatioos”--Bd1. Am. Met.
8 0 C ., vol. 16. Nos. 6-9, August-September 1935.
328 MONTHLY WEATHER REVIEW PEPTBYBER 1937
ft,.; so i t is probable that the 16-inch white balloon assures daytime, clear sky obse,rvo.tions t80 30,000 feet in any ordinary wincls.
In 1937 sonie red 1G-hch bnllooiis were obtained, mid test,ecl aga.inst white balloons to nlt,itiicles around 30,000
fee,t,. Under clear skies the whitlitc bulloon gave a shttrper
iinsge and coulcl be followed to somewhat higher altitudes, t,hough the red balloon proved nearly a,s good. Under
cirriform c,loiicls the red balloon w-as easier to follow, but,
with duo care the white balloon could also be followed to high alti ti ides.
I n connect,ion with development work on the light
sources described below, ascen ts were made wjth 16-inch
balloons carrying various weights. The additmion of any weight up to 5 or 10 oz., hung by a string 50 ft. below t,he
balloon to nlininiize swinging, had much less effect on rise
rate than one might expect; apparently the weight stopped tumbling t,e,ndencies and pulled the balloon into a more
streamlined shape, thereby conipensating for the loss of
free lift. With a weight of 5 oz. or so, indeed, it seemed necessary, on the nvemge, to clecrea.se the froe lift in order to obt.nin the same rise rate in lower and middle levels.
Wibh the weight,, however, t,here was litt,le if any increase
of rise rate w-it.h altitude; a t night, in fact,, t'he ratme usually decreased somewhat wit,h altitude. In general, the rise
mtes with weights a.ppeared to be more uniform a t all times, and suggested the idea (w1iic.h could not be t,hor-
oughly investigst'ed) that dummy weights (say 5 oz. or
so of some light, che.a,p materia.1 siwh as pasteboa.rd)
might profitably be used for bet,ter uniformity in daytime
ascent,s. As an imnie,diat,e and fortunate corollary of the balloon-plus-weight perforinance, it was evident that fairly heavy luminous sources could be ca.rried up at 400
yd./min without. overinflatsing the balloon.
SOURCES OF LIGHT
Six-inch bdloon. li!ghts.-Standard lights for &inch bal-
loons, as used by t,he United Stat'es Weather Bureau, are of
t>wo kinds: A single-c.andle lante,m and R sinal1 electric light. The single-candle lantern consists of a candle about 3 inch (diameter) by 1 inch (long), candlepower about 1.0
:md quite constant, burning time about 35 min. ; mounted in the bottom of a cylindrical paper lantern about
inches (&meter) by 5 inches (long), transmission of
paper about 50 percent,; weight cornplet'e about 34 oz. and cost perhaps 2 cm ts. Under averngc-visibilit,g, clear-
sky condit.ions, this lantern can be followed with a standard
theodolite to c1istmc.rs of 3 to 6 miles, corresponding in various winds to albitudes of 6,000 to perha.ps 15,000
feet, above station a t n rise rate of 200 yd./min. The st,nnd:ird e.lectric light consists of two 1.5-volt dry
aells, size % incA by 17; inches, :tiid n, flashlight bulb ra.ted
a,t 2.3 volt,s, 0.35 ainps.; opm circuit, voltage 3.0; a,t. st'o.rt
t,lie dosed circuit, volt,n.pe is 3.3, the watkage 0.6, and t,he candlepower about 1 ; at8 assumed ciit,-ofT, after about, 30
minute.s, t,he volta.ge is 1.5, the v.-a.tt,a.ge 0.3, and t,he ca~idle-
power about g; the light we.ighs about 1 oz. and cosb perhaps 12 cen t,s. Compared with the standard cmdle Innt,ern, the e.lect,ric light gives c,ompnm.blc altitude per-
formance; it also eliminates fire hazard. Four-candle lan.tern..-Anp sbudy of t'he problem of
ren.ching higher n.1 titudes wit,h night upper wind observa-
tjions n t once siiggest,s brighter light; sources RS the niost,
obvious answer. The most proniismg immedin te np-
proach appeared t,o be an enlargernent of the standard candle lantern. Esperimentat,ion with various sizes and
arrangement's of candles and la.nt8erns, inc.luding some very
large candles n i d e to order, showed that the most practi-
cal development was a multiple nrranpenient of standard candles spaced abou t 2 inches apart in a completely closed
paper 1:mtern Inrgr, enough so as not, to catch fire. To reach distances wid ultitudm twice as great as those
possible with the single candle ltntern, a t lrnst four times
as much li ht would he needed. The developed lantern
Four-candle lantern:
had the fol f owing characteristics:
Cylindrical ahite tiswe paper lantern, 12 inches in diameter by 30 inches long; top and bottom frames, circular with cross pieces, of No. 16 copperclad steel wire soldered together at junctions; hanging wire to balloon string, and safety wire between top and bottom framce, No. 20 soft copper. Fnur-candle holder of %-inch plya ood, 3 3 inches square, having four lHo-inch holes, drilled on diagonals, 1 !& inches from center so aa t o hold four staudard yi-inch candles, each 3 inches long, with wicks at the cornera of a s uare 2 inches on a side; holder secured to bottom frame 07 lantern with thumbtacks. Weight complete: 5% oz. (150 g approximately). Four candles: 2?4 ox. (80 g). Candle holder: 1 02. (30 g).
Lantern: 1% 02. (40 e ). Candlepower: About 4 to G .
Transmission through paper: About 70 to 80 percent. Burning time: Ahout 50 miniites. Four-candle lantern with 16-inch balloon: Free lift: 38% oz. (1,090 a ). Average rise rate: Around 400 gd./min.
R l d e the lantern reaches 30,000 feet in 25 minutes and
is usually lost earlier, the 50-minute burning time allows ample reserve for lighting the candles nnd pasting or clip-
ping closed the bottom of the lantern at leisure, for use a t
a slower rise rate with a 12-inch balloon or underinflated
16-inch balloon, or for observations to extreme altitudes
under very favorable conditions. The lantern is launched
with even acceleration so that it will not jerk out of the hand. The launching location must be sheltered from anything greater than light surface winds, which might blow the side of the lantern into tho flame. Once released,
the lantern almost invariably stays lighted despite the
terrific beating i t takes from the combined forces of down- draft, humping, and swinging. One lantern, in fact, re-
mained lighted fc,r over a minute in free fall after its bal- looii h r s t a t high altitude. As the candles burn down, :I
pool of was forms on m d around the cnnille holder; this finally ignites and burns away the lower part of the lan-
tern; the air draft then extinguishes the flxnies entirely,
so that, short of premature balloon bursts, the fire hnxnrd is not serious.
Foiir-candle lanterns were used in about 60 test ascents, of which :ihout 30 were night douhle-theodolite runs using
16-inch balloons. The large lantern presents a largo
lighted surface easily focused into a clear image. Several tests sho~eed that it, can be seen two or three times as far
away as the single-cnndle lantern: nncl rising a t 400
yd./niin., it gives observations to altitudes two or three times 0 3 great RS those possible with a single-candle lantern
rising a t 301) vd hiin. Our best night observation uith
the foiir-candh lantern eytended to 35,501) feet altitude
above station. checked by double tlieodolite on the 20th minute when the Inntern was about 13 miles distant and encoiintering 70-niile winds. A sinple-candle lantern re-
lensed within the hour, under identical conditions, could
he followed only to 8.000 feet altitude, where it was lost in 30-mile xincls. On n ninter night when winds abore 10,000 feet were 60 to 90 ni. p. h ., four-cnndle lanterns
were followed t o 15,000 feet altitude, ns against 7,000 feet
for single-candle lanterns. All these trsts, a3 well as the tests described hereafter, were made near the east coast
where viqibility is rarely rery good; in the clear air of the
I
L
i
t
Volume 65, No. 9 Monthly Weather Review, September 1937
FIGURE 3.-Right: Four-candle unit as used in four-candle lantern. Left: An experimental eight-candle unit. Size comparison: 16-inch slide rule.
FIGWEE I.-Left: 3-watt B-oz. electric light 8 0 ~1 ~8 giving night upper wind observations to altitudes of 16,OOO to 25,000 feat. Right: Acstylene light source.
IB
SEPTEMBER 1937 MONTHLY WEATHER REVIEW 329
West, bet,ter results might be expected. Summariziug, we might say that the four-candle lantern can be followed
by standard t,heodolites under favorable conditions to dis- tAnces of 8 to 15 miles, corresponding in various winds to
altitudes of 15,000 to 30,000 feet at a rise rate of 400
yd. Imin.
Six- an.d 8-candle 1anterna.-In an effort to better
the performance of the four-candle lantern and explore the limits of multiple-c,andle combinations, eight-candle and
six-candle units were developed for use with the 12-inch b 30-inch paper lantern described above, which in itse 9 f re resented a desirable limit of overall size.
\he eight-candle holder consisted of ti plywood ring drilled with eight %-inch holes a t 1%-inch intervals around
a 4-inch circle, in which eight 'X2-inc,h votive candles could
be fixed. Candles 3% inches long gave a lantern burning
time of about 25 ininutes and a total weight of about 594 oz. ;
with 5-inch c.andles it burned about 35 minutes and weighed
about 7% oz. This lantern gave some increase of light over the four-candle type, but sac.rificed too much in simplicity and reliability. It caught fire easily in light surfac.e winds,
and in some balloon tests we.nt out at middle altitudes.
The six-candle lantern was develope,d to give somewhat more light than the four-candle lantern without sacrific,ing
simplicity or reliability. To keep the weight within reasonable limits, however, it was necessary to sacrifice
reserve burning time. Summarizing approximate char- acteristics:
Six-candle lantern :
Cylindrical white tissue paper lantern, as described above. Six-candle holder of %-inch plywood, 5g-inch ring with 2,g-inch center hole, having six 1%&nch holes drilled at equal intervals around a 4-inch circle so as to hold six standard %-inch candles each 2 inches long with wicks spaced 2 inches apart around the circIe. Weight complete: 6 5 01. (185 g).
6 candles: 3 01. (85 g ).
Candle holder: 2 08. (55 8).
Candlepower: about 6 to 8 (lese 20 to 30%). Burning time: about 25 minutes.
Free lift: 40 oa. (1,130 g).
Average rise rate: Around 400 yd./min.
Six-candle lantern with 16-inch balloon:
In addition to ground tests, only three double-theodolite
night balloon ascents could be made with the six-candle lantern in the time available. In one comparative test
under fair conditions the six-ca.ndle lantern was followed
to 19,000 feet altitude as ttgninst 16,000 feet for the four- cnndle lantern. On another comparative test, both went to about 24,000 feet. By saving weight on the candle
1iolde.r and lengthening the candles somewhat, t'he burning
time of t,he six-candle lantern could perhaps be estendcd to 40 minutes, keeping the total weight under 7 oz. In its
present form, the six-candle lantern appears to give slightly
better performance than the four-aandle lantern, part,icu-
lady when visibility is only fair. Another promising development that could not be com-
pletely tested was the four eight-candle lantern, which
offered the very desirable qua.lit;v of incre.asing candle-
powe,r. Characteiistics:
Four eight-candle lantern: Paper lantern as described above. Candle holder: Same as itvfoti'r-candle lantern. Four y&inch candles, each 2% inches long, are set in the four holes, same as in four-candle lantern. In nddi- tion, four ?/-inch candles, each 1% inches long, are set on the diagonals just inside of, and tangent to, the outer candles. Only outer candles are lighted at start; inner
candles ignite from them after about 20 minutes. Weight ccbmplete: About 6% 02. (180 g) . Candlepower: About 4 to 10. Burning time: Around 25 minutes. 400 yd./min. rise
rate with 16-iqc4 balloon given free lift of about 40 OS.
(1,130 B).
No air tests could be mnde with this lantern in the avail-
able time. Ground tests (of candle unit in open air) showed candlepower about 4 a t start, increasing to 5 and 6
within 15 to 20 minutes (due to mutual heating of four
candles). A t 20 minutes the first inner candle ignited; the second ignited a t 21 minutes, raising the total candle-
power to 8; the third lit nt 23 minutes; the fourth lit a t 25 minutes, raising the total candlepower to 9-10; at 28
minutes, two candles had burned down to the wood
holder. \Illether the inner candles would ignite properly
in the lantern during a free asecnt, is not known; it is believed that some of them would, a t least.
The large candle lanterns described above, compared
with other possible light sources of comparable power,
have several advantages: Universal availability of mate-
rials, low cost (about 6 cents), harmlessness on descent,
and reliability. Of the three flame-type light sources that
we tested, the candle lanterns alone remained lighted to altitudes above 20,000 feet. All the multiple-candle
lanterns, also, have the desirable quality of increasing
candlepower in some degree. The disadvantages of large candle lanterns are: Slight fire hazard, limited candle- power, hnndling difficulties unless carefully folded, and
launching difficulties in high surface winds.
Three-wutt electric light.-In order to obtain an electric light source about four times as powerful as the standard
&inch balloon light, we tested various combinations of
batteries and bulbs. For simplicity, only standard units
in wide distribution were considered. It was early ap- parent that any electric light of this type would leave much
to be desired in the direction of light weight and sustained Candlepower. The best combination appeared to be as follows:
Three-watt electric light: Battery: Sis 1.5-volt dry cells, size 1 inch by 1% inches; six &volt auto bulb, rated at 3 watts or 1.5 candlepower; weight complete about 9 01. (255 g); cost about 40 cents. Data a t start: Open circuit voltage 9.0, closed circuit voltage 7.5, watts 3.4, candlepower about 4 to 5. Time at assumed cut-oiT 45 minutes; closed circuit voltage 4.5, watts 1.5, candlepower about two-thirds. Candlepower at 25 minutes (30,000 feet altitude) : about 1.
Free lift: 42 oz. (1,190 g). Average rise rate: Around 400 yd./min.
Three-watt electric light nith 16-inch balloon:
In addition to ground tests, five double-theodolite night balloon ascents were made with this 3-watt electric
light. Just before hunching, the wire leads soldered to
battery and bulb, scraped bright, were twisted together.
During the first few minutes of any ascent, while the bulb
operated a t high efficiency above its normal voltage, the light was brilliant blue-white, and so steady thak it showed
in the theodolite considerably better than its measured
candlepower would indicate. Later in. the ascent, as the
bulb efficiency declined, performance was about equal to,
or slightly less than, that of the four-candle lantern.
The best night altitude reached was 23,250 feet, when the
light wns about 10 miles away in 70-mile wincls 18 minutes
after leaving the ground.
The main disadvantages of the electric light are: Its
considerable weight, considerable cost, and its very
limited and steadily decreasing candlepower. Weight might be saved, and candlepower held nearly constant, by using specially built batteries of lead-acid type. The
9-ox. unit, fnlling from several miles altitude when the
balloon bursts, would be a hazard to objects on the ground; for this reason a simple parachute, consisting of
a cloth 3 feet square, was connected in the string between
balloon and light. This parachute functioned satis-
factorily when tested in tbe daytime under a balloon
330 MONTHLY WEATHER REVIEW SEPTEMBER 1937
carrying a small bursting charge of blhck powder attached
to the neck, the powder being i nited after a,bout 5 minutes
cells lose their effic.iency very quickly a t t.empera.t,ures
below about 5' F., some sort of heat-insulating materinl, such as several thicknesses of paper, has to be placed
around the batte.ry. The advantages of the electric
light are: Ease of handling, ease of launching regardless
of surface wind, steadiness and dependability of the light itself, and absence of fire hazard.
Bcetylen4 light.-This light was developed by A. P. Rehbock. It, coneists of a 6-hich balloon inflated with
acetylene, feeding a V-type burner protected by a conical, aluminum reflector-windscreen. The acetylene balloon hangs, burner downward, by a string from the main
balloon; inflated to 12-inch diameter, the small balloon
supplies the burner for about 35 minutes; the candle-
power is around 20. The total weight is about 1 oz;
cost, about 10 cents. Acetylene lights were tested in five night-balloon ascents, two a t a 200 yd./min. rise rate (16-hch balloon;
free liiht 10 oz.), and three a t a 400 yd./min. rise rate (16-inch balloon; free lift 38 oz.). At 200 yd.pnin. the
light performed well-certainly much better than a single-candle lantern or small electric light of compara-
tive weight. But the acetylene light flickered badly (at
a period of about one-hnlf second), a t times disappearing altogether for a moment. Each fade and come-back was
so rapid, however, that the light could be easily followed
with a theodolite. At 400 yd./min. the acet'ylene light went out in all
three tests, one a t a few- hundred feet, one a t 7,000 feet, and one a t 14,000 fe,et. From the,se incomplete d a h , it
would appear that the present form of acetylene light is
not reliable a t the high rise rates which are nec,essary to reach high altitude,s in high winds. Perhaps changes in
design, such as an improved reflector-windscreen, would remove this defect. The acetylene light's advantages are
light weight, low c,ost, considerable candlepower; it,s dis- advantages are difficulty of handling, inflation, etc.,
flic,kering, and (at present) unreliabilitg.
Pyrotechnic fare.-& the Strntoc,azlip in June 1935,
while experimenting with 12-inch balloons and four-candle lanterns, we made several night ba.lloon tests wit,h small,
standard railway flares. The %-inch, 10-minute flare weighs about 5 oz., and burns with a brilliant red light of more than 200 candlepower; our night balloon tests
showed it to be far superior to any other light source during it,s short burning time. I n sphe of some fading,
the light w-as so powerful that it could be easily picked up with the naked eye a t a distance of 3 to 5 miles. The
problem was to develop a. red flare t,hat weighs less t,hnii
10 oz., arid burns at least 30 n h u t e s under a balloon rising
a t 400 yd./min.! with cmdlepow-er a t least 50 a t the start and preferably increa.sing gradually t,o several hundred as
the balloon rises. The problem is still unsolved, despite
considerable work by esperts in the fireworks field, but
considerable progress hns been made. Other pyrotechnic lights, suc.11 as magnesium ribbon,
magnesium flares, and sparkler matmerial were considered, experimented wit,h, ancl dropped in favor of the more
promising red flare; red light penetrates haze best, and is
easily distinguished from stars. Over n period of more
than a year, samples of four flare models were submit,ted,
each in some respects an iniprove,ment on the prec.eding
one. The first flare model embodied a flare tube around 50
inches long; its small (initial) end diameter was about
Ks inch (the smallest diame,ter t,hat would burn evenly
of ascent by about 6 feet of E lasting fuse. Also, as dry
and reliably in still air), increasing coiist8antlp to about $-inc,h diameter a t t,he large (terminal) end; the weight
was about 7 oz. The flare tubing w:is arranged in spiral form on a light, ineta.1 frame, to burn inwnrds from the
small end to the large end in about 40 minutes, with cmdlepower increasing gradually from about 30 to about
100. Tested on the ground, t,his flare burned satis- fact,orily esce.pt for severe dimming (perhaps due to
conductive, loss of heat) where the burning end of the tube
passed t8he metal fra,nie members. In about six night balloon te,sts this dininling was very appnrent, though t,he light usually brighkned aga.in be,fore it was lost to the
theodolite. The best asc,ent reached an nltit8ude of nbout
14,000 feet. In some ways this flare, with ibs tube hung
horizontally, peipendicular bo the slipstre,am, performed
better than later models that, were hung vertica.lly. Some of our later flares performed sat,isfa.ctorily under fan down-
drafts up to 20 m p. h., yet we,nt, out in air tests. The wriggling, swinging, humping motion of the balloon rising
a t 400 yd./min. was something, apparent,ly, that could not, be duplicated by any tests on the ground
The second flare model, to obvia.te the dimming noted nbove and to simplify construction, was a straight, tapered
tube. It was 25 inches long; the diameter was three- sixteent.hs inch at the lower (initial) end nnd t,hirteen-six-
t,eent,hs inch a t the upper end, whence a wire connect,ed to the balloon string; the weight was 7 oz. and the burnjng t,inie (horizontal in still air) 45 minut,es; t'he cn.ndlepower
increased graduallp from about, 30 st the start to a.round
150 a t t,he end. About sis night balloon tests showed that some fading had been elirninat,etl, but considerable re- mained. All these fla,re.s blew out a t altitudes below 10,000
feet, probably because the composition was t.oo slow- burning and the initial dia.meter too small. Although the idea of a t,apered tube and increming cnncllepower wns a.bantloned in sucweding models in order t,o simplify cnn-
struction and reduce cost,, it is be,lieved to be worth further
conside,ration.
The third aud fourth flare models were straight, uniforni cylindrical tubes, similar to railway flares but t,hinner and longer. Summarizing t,heir chs.rac.teristics:
No. 4 uvrotechnic balloon flare:
Si&: Weight: 9% 02. (270 g).
Burninx time: About 30 minutes.
inch (diameter) by 24 inrheR (length).
Candlepower: Average about 100. Candlepower 'for last minute: More than 500. Flare with 16-inch balloon: Free lift: 43 02. (1,190 g). Average rise rate: -4round 400 yd./min., variable. 16-inch safety balloon inflated to 10 08.
Tests with the model 3 flares showed considerable fRd-
ing, appareii tly cnused by temporary accumulations of nsh from the pasteboard container tube as it burned away.
These ash cones would stifle the flame somewhat for
several seconds, then drop away. Consequently a faster- burning composition R ~S used in niodel No. 4 (necessarily
cutting clown the buniing time), with improved results.
Foiirteen night-bnlloon tests were made with model 4
flares. They showed some fading, inseparable from any
light source of this type; but the fading wns not serious, tind at) lower levels the light was far brigliter and clearer
than any other balloon light, being visible for several minutes to the naked eye as a red, olwbright star. The
best altitude checked by double theodolite was 17,280
feet, when the light was about 6 niiles distant in 17-mile winds. It disappeared suddenly from satisfactory bril-
liance. Ten of the flares went out similarly at altitudes
between 9,000 and 16.000 feet. It is possible that lack
of oxygen, which would be apparent o t these altitudes,
SEPTBYBER 1937 MONTHLY WEATHER REVIEW 331
was the cause; though the composit,ion includes its own oxygen, atmospheric oyygen niight be necessaiy to the
steady burning awa,y of t,he past.eboard.
From the results de.scribed above, it appears that a still better flare might be built on the following specificat,ions, using a fairly quick-burning composition: Tube, of diame-
ter increasing from about eleven-sixteen ths inch a t lower (initial) end to about seven-eighths inch a.t upper end;
candlepowe,r gradually inc,reasing from about 100 a t start
to about 500 a t end; burning time, 25 minutes; weight, less than 12 oz. This desirable development, however, could not be undertaken in the t8ime available.
The disadvantages of the flare are its unreliability (at
present), its fairly heavy weight, and its se,rious fire hazard. I n one or two of our tests, flares dropped several
thousand feet due to wire breakage or premature balloon
burst, and remained lighted all t,he way to the ground. We therefore used a strong (No. 16 copperclad steel)
safety wire, and also a safety balloon on each of tlie h t e r ascents. The function of the safety balloon, a 16-inch
balloon inflated to 10 oz. free lift, was to hold the flare off
the ground in case the lifting balloon burst; for least osc,il- lation and most uniform ascent rabe, the two balloons are preferably tied neck to neck. The advantages of the
flare are its simplicity, ease of launching, and its out,st,and-
ingly bright light when operating properly.
OBSERVATIONAL TECHNIQUE
Theodolite considerations.-All the tests described above
were made with the standard theodolites in common use for upper wind observations. The observation of a
planet or two (such as Jupiter and Saturn) m-ith this instru-
ment reveals that, optically, it leaves much to be desired.
For high altitude night observations a larger objective- say 60 mm-would be far more satisfactory due to its
increased light-gathering power and better definition.
Two or three eyepieces, quickly interchangeable, say 1 OS, 20X, and 40X, should be provided. In daytime obserra-
t.ions under a clear sky, a red filter can often be used to
advantage, the balloon appearing as a bright red speck against a dark background. Vl-hen trymg for extreme
altitude under unfavorable conditions (far out, the balloon
appears as a very faint and evanescent speck), each theo- dolite should be manned by two men-one to follow the
balloon and the other to read the scales. For night work the reticule should be lighted interiially and the scales lighted exteniully; the brightness of each electric bulb being controlled by a separate rheostat.
Above 10,000 feet, under most conditions, two men sbould be on each theodolite. The theodolite must be screened
from nearby surface lights that bother the observers. SinglP versus double theodol:te.-The average rise rates of 6-inch balloons have been thoroughly studied.' For-
mulas have been developed giving rise rate in terms of balloon weight and free or total lift; with the small balloons these formulas are reasonably accurate. With the stand-
ord free lift, 4.66 oz. (137 g), a 6-inch balloon usually (hut not always) ascends a t a rate within *l q percent, of 200
yd./min. Most of any series of nscents wdl be within the
& 10 percent limits, but occasionally a balloon will depttrt widely from the standard rate, perhaps by ns much as
50 percent. It is apparent, therefore, that a one single-
theodolite balloon ascent cannot, be trusted blindly for accuracy (if accuracy in a gjven case is important), al- though in general the upper umds are given by single theo- dolite accurately enough for present-day uses. There is
4 B. J. Sherry. The Rate of Ascent of Pilot Balloons. M. W. R. vol 48. Decernber
I W. C. IIalneb. Ascenslonal Rate of Pilot Balloons. M. W. R. vol SI. May 1924. 1920.
in aviation, however, some demand for more dependable upper winds espressed in magnetic, degrees of azimuth.
Our tests with the 12-inch balloon showed that, in general, with tl single theodolite, it w-ould give results about as good tis the 6-inch balloon. For dependable upper wind observations to high altitudes Nith the 16-
inch balloon (and no smaller balloon dependably reaches
these altitudes), our tests shou-ed double-theodolite ob- servations to be very desirable.
Double-theodolite observations metin, in optimum practice, t h e e theodolite positions loca
alternate base lines roughly at right angles. For best results nll three theodolites should be manned, the base
line being chnnged tis necessitated by the balloon drift. Even for high-altitude observations, %mile base lines
usually suffice. All theodolite stations should preferably be connected to the plotting hoard by dependable wire telephone. With a well deaigi)c.cl plotting board, double or triple
theodolite deterniina tion of upper wincls is simple and
easy. The bonrd used in most of our tests was devel- oped as the result of several years' esperience nt Aberdeen Proving Ground. The fixed center pin represents the bal- loon. From this pin radiate threc. mms (set with refer- ence to :t large degree circle on the board) representing
azimuth froin station A, nziniuth from station B, and
elevation from station A. The station A azimuth arm is prorided with a perpendicular altitude m n (set against the station A elevation arm). Any desired base line can be chosen from a four-theodolite network; on the board a base line (or double or qundruple base line) is represented
by a sniall ruler-like straight edge clamped to a drafting machine which holds it always parallel to its actual azi-
muth wherever it is moved on the board. Base lines can be switched during an ascent, and velocities of upper winds are determined as the bnlloon rises to new levels
without any delay.
High-altitude upper-wind po.ssibilities.-Regardless of
aviation or other demand, it would not seem possible to put in high-altitude ftjcilities a t each of the eiglity-odd
upper-wind stations now operated by Weather Bureau, military and nnval, and private services in the United
States. One might visualize, rather, a limited network of
15 or 20 high-altitude, upper-wind stations superposed on
the denser low-altitude network. The high-altitude upper, w-ind stations, indeed, might coincide with the present airplane-sounding stations, which should change gradually to radio-sounding-bnlloon stfttions in ally case. Eren- tually, when better solutions are found to the difficult
radio-pilot-belloon problem, the same stations could be
used for this purpose also. Thus 15 or 20 high-altitude upper-air stations scattered over the United States might
eventually determine upper winds to 30,000 feet altitude
by balloons and light sources similar to those described
above, under clear skies or ciiTiform clouds; they niight
determine upper winds by radio-pilot bidloon through
lower clouds or poor visibility ; and they might determine air-mass clmrac teristic s from pressure, , temperature, and
humidity values given by radio-sounding balloons.
ACKNOWLEDGMENTS
The writer aclmowledycs with gratitude the assistance
of the following noncomniissioiied officers of the Army
Meteorological Service in this research: €1. J. Pryber, R. ht. Glenn, L. B. Burke, W. F. Bernheisel, F. A. hhtchin- ski. Also gratefiilly acknowledged is the cooperation of
Lloyd A. Stevens of the United States Weather Bureau,
m d Cnpt. D. W. Vl'tttkins and Lieut. F. B. Wood, Air
Corps, United S ~H tes Army.