SHORT PERIOD VARIATIONS IN RADIO RECEPTION*

GREENLEAF W. PICKARD

(CONSULTING ENGINEER, THE WIRELESS SPECIALTY APPARATUS COMPANYBOSTON, MASSACHUSETTS)

(Communication from the International Union of ScientificRadio Telegraphy)

In any continuous measurement of the field intensity from adistant transmitter, we find both periodic and irregular varia-tions, often of great amplitude, and always of complex form.Some of these changes in the transmission coefficient of the radiocircuit may be definitely correlated with solar and terrestrialhappenings; for example, transmission is in general better in winterthan in summer, better at night than during the day, and is usu-ally adversely affected at the time of intense aurora. These fluc-tuations are less prominent at the lower transmission frequencies,and increase until several thousand kilocycles is reached. Then,altho the data at our disposal today is quite limited, they appearto decrease, until at ten thousand kilocycles, more or less, thetransmission becomes quite uniform.' Altho there is little doubtbut that these fluctuations are due to changes occurring some-where in our atmosphere, we are to-day in ignorance as to themechanism involved.

If we confine our attention to frequencies between five hun-dred and fifteen hundred kilocycles, that is, to the band nowprincipally filled with radiophone broadcasting, and to overlandtransmission at distances between one hundred and one thousandkilometers (62 to 620 miles), we find that the average midwinterfield intensity is about five times greater than in midsummer.

In the daytime, altho transmission is relatively free from thelarge amplitude short period variations commonly known as"fading," "swinging" or "soaring," there is usually a slow changefrom hour to hour, which in the majority of cases appears as a

*Received by the Editor, November 8, 1923. Presented before TlHEINSTITUTE OF RADIO ENGINEERS, New York, Dece-mber 12, 1923.

1 "Radio Telegraphy," Marconi, PROCEEDINGS OF THE INSTITUTE OFRADIO ENGINEERS, volume 10, August, 1922, pages 235 to 236.

119

gradual decrease from morning to night, the morning intensitiesoften being twice those of the afternoon.

About half an hour before sunset at the receiving point, theweak and relatively constant field from the distant station beginsto show marked short period fluctuations which grow in ampli-tude from minute to minute until, soon after suinet, they usuallyassume grotesquely large amplitudes. The principal changefrom day to night is an increase in field intensity; in general, thelower limit of the night-time field is approximately the same asthe late afternoon field, altho from time to time there will befound a nmomentary fall to a rnuch lower value than at any timeduring the day. The upper limit of the night-time field is notso definite; it may be ten, a hundred or even on occasion thous-ands of times greater than the daytime intensity, depending uponthe distance and the character of night.

The amplitude of the short period variations, that is, thosefluctuations ranging in duration from seconds to tens of minutes,is principally controlled by the distance between the transmitterand receiver, and this is true of both night and day transmission.At less than eleven kilometers (6.9 miles) from a broadcastingstation, there is a well-defined short period fluctuation in inten-sity during night-time transmission, which is practically absentduring the day, and which on some evenings shows an amplitudeof ten percent or over. As the distance increases, the amplitudeof the variations also increases, becoming readily observable byday at distances of fifty kilometers (33 miles), more or less.

At first there is no change in the character of the fluctuationother than in amplitude, but when a distance of between one andtwo hundred kilometers (63 and 125 miles) is exceeded, the oscil-lations of periods ranging from seconds to a minute or two becomeless prominent, and the longer swings of minutes to tens of min-utes are accentuated. At an ill-defined distance of perhaps onehundred and fifty kilometers (94 miles) the amplitude of theshorter period variations appears to be at' a maximum.

When continuous records of night-time field intensity froma distant station are made at separated receiving points, I havefound that altho the records are of the same general characterthey are not identical unless the receiving points are practicallycoincident. A separation of less than six hundred meters(1,970 feet) between two receiving points is sufficient to obliter-ate most of the detail resemblance between their records, saveonly for the more prominent long period swings. Equally,simultaneous records at a single receiving point from separated

120

transmitters show few detail resemblances other than the sea-sonal and diurnal changes, altho here there is, of course, a differ-ence in frequency.

A comparison of records made on different evenings shows thatthese evenings may be roughly classified as quiet, moderatelydisturbed, or highly disturbed for all stations within a two orthree hundred kilometer (125 or 188 mile) radius. Whenever astation eleven kilometers (6.9 miles) distant shows marked shortperiod fluctuations, more distant stations will also exhibit thesame disturbances. On a quiet night the local station shows littleor no variation in field intensity, and the more distant stationsshow principally long period disturbances.

The records of broadcast transmission which form the prin-cipal part of this paper have all been made by reception on non-directional open antennas, as my object here is to present only truechanges in intensity of electric field, as distinct from the changesin wave-front which often accompany them. In a subsequentpaper I intend to present records made on both stationary androtating directional aerials, in which I have found striking changesin wave-front over very short intervals, some of these changesbeing very definitely correlated with changes in true field in-tensity.

The received current in the open antenna is impressed upona crystal detector, directly if the field intensity is sufficient, butusually after from one to four stages of radio frequency amplifica-tion, and the rectified current from the detector is then passed thrua galvanometer. In one form of recorder employed for makingthese records, the registration is photographic, a beam of lightreflected from the galvanometer mirror playing across a sheet ofphotographic paper wound on a slowly rotating drum, while inanother and more portable form of recorder a pointer galvan-ometer is used, the registration being semi-manual.

One of the arrangements employed for recording is shown inFigure 1. The open antenna 0 includes a filter F, for minimizinginterference from local stations, a series tuning condenser C 1,a loading inductance L 1, and a coupling coil L 2. A groundedshield S electrostatically separates the antenna circuit from thetuned grid circuit L 3, C 2 of the radio frequency amplifier train.The amplifier circuit here shown is that of Mr. Chester W. Rice,2which I have found very stable and constant, and in every waysuitable for this rather exacting work. The voltmeters VM 1and VM 2, which measure the filament and plate voltages in the

2 United States patent, number 1,334,118.121

amplifier train, are essential adjuncts, as it is of first importancethat the voltage amplification be kept rigidly constant over longperiods.

The amplified radio frequency current res-alting from this trainis impressed upon a crystal detector D, and the rectified currentfrom this detector then passes thru a reflecting galvanometer RG,provided with a constant impedance shunt CIS for adjusting thedeflection to the width of the record sheet. Monitoring tele-phones MP are shunted across the galvanometer circuit, with aseries condenser C 8 to prevent any drain of direct current fromthe galvanometer circuit. A light projector P 1, consisting of asix-volt 2-candle-power lamp, a pinhole diafram, and a lens,throws a beam of light upon the galvanometer mirror, fromwhence it is reflected to a sheet of photographic paper woundon a kymograph drum K, which is normally rotated once an hourby a clockwork motor M. A similar light projectoi P 2, provided

o

C's

CI AL

s ~~~~~~~~~~~~MP

L2 z z CG C77- Cs -~~Cs L L

B_ O~~~~~~~~~~~

FIGURE 1

in addition with an astigmatic lens A L, flashes a short line oflight once every minute on the lower edge of the sensitive paper,thus giving a time scale.- The galvanometer employed is of fairsensitiveness, giving approximately one millimeter deflection atone meter for a current of 10-9 arnpere, so that it is not necessaryto overload the crystal detector, and so depart from the current-square relation between input and output. The kymograph,galvanometer, and light projectors are, of course, installed in a

dark room, with shielded leads running in from the detector,which with its associated amplifier, tuner, and filter is placed in

122

another room. A crystal detector preceded by a radio fre-quency amplifier train is extraordinarily stable in adjustment,and the one employed at my receiving station in Newton Centre,Massachusetts,* has operated without readjustment for over fivemonths at a time, with no measurable change in sensitiveness.

Another form of recorder which has been extensively em-ployed in this work, and which has the advantage over thephotographic form of portability, freedom from photographicprocess and delay, and great simplicity, is due to Mr. H. S. Shaw,Jr., who has been associated with me in this work almost fromits beginning. A photograph of this recorder is shown in Figure2. A strip of paper, 9 cm. (3.55 in.) wide, passes around a drum

LV

'l AIP0D

&VM

FIGURE 2

D, rotated by a worm and gear drive from a Warren ClockMotor W M. A pointer galvanometer G to the left of the drumis connected to the crystal detector, and movements of its pointerare followed manually by the index I (the shadow of which is

123

cast on the galvanometer scale by the light L) which in turntransmits the motion to an inking pen IP, traversing the paperstrip. It might be thought that this form of recorder would berather tedious to operate, but after a few minutes' recording themovements become almost automatic, and a record an hour longis not at all tiresome. The time scale of this manual recorderis made more open than the photographic, being 1.6 cm. (0.63 in.)per minute, while the kymograph scale is only 0.5 cm. (0.19 in.)per minute.

With a fair intensity of electric field, that is, of the order of athousand microvolts per meter, and a fair sized open antenna,excellent records may be made without any amplification at all,direct from the crystal alone, as may be seen from Figure 3.recorded by Mr. Shaw at Scituate, Massachusetts, from Schenec-tady, New York, 265 kilometers (165 miles) away.

KDI2PMEFJ /312 <

FIGURE 3

Another almost equally simple arrangement is to shunt thecrystal detector across the plate impedance of a regenerativetube, and with this good records may be made with electric fieldsof a few hundred microvolts per meter.

As mentioned above, the ordinates of these records are veryclosely proportional to the square of the electric field, so that theymight properly be termed energy records. For many of theserecords I have determined approximately the values of theordinates in microvolts per meter, and I have used this data indetermining the seasonal variation in transmission. Owing tothe extraordinary complexity of many of these records, it isdifficult to determine the mean value over any extended periodby mere inspection, and the method which I have adopted is tomeasure the area of the space between the curve and its base orzero line with a polar planimeter, and divide the result by thelength of the base line.

*Newton Centre is 11 kilometers (7 miles) in a westerly direction fromBoston, Massachusetts.

124

Altho the methods outlined above appear and are actuallysimple, it is not altogether easy to obtain an unbroken series ofgood records from distant stations. When the field from thedistant station is weak, high amplification is necessary, and theoutfit is then at the mercy of disturbances, such as local sparkstation, nearby amateur working, and even a violently oscillatingreceiver in the immediate neighborhood. Also, the record maystart at a time of low field intensity, and the galvanometer shuntand amplification may be so pessimistically chosen that when theintensity rises most of the record will be "off scale."

Before taking up the records in detail, I wish to make clearthe relation between the intensities shown on my records andthose observed by ear, either from phones on the head or froma loud speaker. In general, the human ear is unable to discrim-inate between sounds of the same character, but of different in-tensities, unless the difference is quite large. For example, trainedobservers making comparisons by ear of telephone transmissionover different circuits are usually unable definitely to determinedifferences of less than one mile of standard cable, that is, achange in amplitude of about ten percent. Such a comparisonis, of course, made under the most favorable conditions possible,that is, the two transmissions are rapidly alternated by switch-ing from one circuit to the other. Where there is no standard ofcomparison, and the observer is listening to a single transmissionwhich is slowly varying in intensity, the change to be observablevaries from about thirty to several hundred percent, dependingprincipally on the rate of change and the absolute intensity of thesound in the ear. If the intensity is fluctuating rapidly, that is,every few seconds, discrimination is, of course, at its best, whereasif the change is a smooth one over a period of minutes, it will passundetected unless quite large. Also, if the sound is very loud,discrimination is very poor. Finally, in listening to transmissionfrom a broadcasting station, the relatively large changes in modu-lation due to the rapidily changing intensity of the voice or musictend to mask differences due to changes in transmission.

In Figure 4 is shown a simultaneous record of the radio andaudio frequency amplitudes in musical selections transmitted fromWBZ, Springfield, Massachusetts, to Newton Centre, Massa-chusetts, a distance of 118 kilometers (74 miles). The fluctua-tions in intensity shown in the lower or radio frequency recordare practically entirely caused by changes in transmission, where-as the variations shown in the upper or audio frequency recordare due to both changes in transmission and in modulation. Thus,

125

from 8.33 to 8.35 P. M. the Springfield transmitter was not beingmodulated, and the audio frequency record remains near its baseor zero line. But over this same period the intensity of theelectric field is fairly high, as is shown by the lower record. In-asmuch as the galvanometer employed integrates the rectifiedradio frequency current over periods of several seconds, and asgood broadcasting stations aie rarely over-modulated, the radiofrequency record, if taken in the immediate neighborhood of thestation, would be practically a straight horizontal line, as willappear from the next figure.

SOOPM 820 840FIGURE 4-WBZ, Springfield, Mlassachusetts Newton Centre, Massachusetts,

118 km. (74 miles), April 6, 1923

The first records I shall show are of daytime transmission.In Figure 5 the distance is 11 kilometers (6.9 miles), and theStation was then operating at 830 kilocycles (360 meters). The

0

12.,IrA Jo 35 40 4S5 So.ooh

FIGURE 5-WNAC, Boston, Massachusetts Newton Centre, Massachusetts,11 km. (6.9 miles), March 1, 1923

more pronounced ripples on this record are undoubtedly due toaccidental variations at the transmitter, such as fluctuations inplate voltage due to power line changes, and possibly to occasional

126

slight over-modulation. If the broadcasting station radiationmeter were of the recording type, it would probably make a verysimilar record. As to the tiny ripples, which occur several to theminute, I am inclined to attribute some of these to true daytimetransmission change, in view of the fact that when the distanceis increased, these same ripples appear with increased amplitude,but without material change in character.

In Figure 6 is shown a superposed pair of records made in day-time from the Boston station WNAC at 1,080 kilocycles (278meters) at two receiving points in Newton Centre, separated 550meters (1,800 feet). The coincidence is practically perfect, whichwould indicate that the fluctuations shown are either all at thetransmitter, or else that they cover a sufficiently large area toinclude both receiving points. The method employed for insur-ing coincidence in time was the very simple and effective one ofmaking a sharp jog in the record line whenever the announcersaid "WNAC."

r

6 MIMES 45 SECONDS -

FIGURE 61

This eliminated any possible difference between the watchesof the two observers, and enabled the records to be very accu-rately superposed.

In Figure 7 are shown daytime records from three distantbroadcasting stations, at ranges of 60, 290 and 225 kilometers(38, 181, and 141 miles), respectively. These records are typicalof average daytime transmission from these distances on moder-ately disturbed days. The points marked "s s s " on theserecords are disturbances due to spark working somewhere in thevicinity of Boston, which are, of course, prominent with therather high amplification necessary for distant day records.

Daytime transmission is, however, different for the samestation from day to day, and this is well illustrated in Figure 8,which is of day transmission from WBZ, Springfield, Massachu-setts, to Newton Centre, Massachusetts, a distance of 118 kilo-meters (74 miles). June 9 was clearly a highly disturbed day

127

I.il.A

WTAR A 23 |

10-.20A-M- 10 40 1100

, Assii U 's '

WJYjJ NE 10. 19p -- _ ycYUNE 10 1923

230PM. 250 950AN. 10 10FIGURE 7 Daytime Reception at Newton Center, Massachusetts, fromWJAR, Providence, Rhode Island, 60 km. (38 miles), WXJY, New York City.290 km. (181 miles), and from WGY, Schenectady, New York, 225 km.

(141 miles)

for radio transmission, June 10 moderate.y disturbed, whileJune 16 was obviously quiet. These three records fairly representthe average and extreme cases of transmission over this particularrange, and altho taken in summer, do not differ from winter day-time transmission.

The normal transition from day to night conditions is wellshown in Figure 9, which covers a period of four hours. The lateafternoGn record beginning in the upper left-hand corner indi-cates a highly disturbed day (it merely appears flat by compar-ison with Figure 8 because of its much lower ordinates) with theusual weak field up to about forty minutes before sunset.

Then well-marked large amplitude fluctuations commence,increasing in intensity, and coming in groups with an interval ofabout 7 minutes. No special disturbance appears to mark exactsunset at either Newton Centre or Springfield, but the amplitude

128

JUNF 9S. 1923. .4 I.

#A04~ W\

IL III I I II IIII L I IIII II I I I I II I I I I I I lI l2-30PM. 2-50

JUNE 10 192-

WFT II I 1"1 I II

3-20 PM . 3 40FIGURE 8 WBZ, Springfield, Massachusetts Newton Centre, Massachus-

etts, 118 km. (74 miles), Daytime

of the short period fluctuations, and also the mean intensity ofthe electric field, increases to a maximum about an hour aftersunset, and then declines again. This record is typical of manywhich I have obtained thru the sunset period, and the tendencyof the short period oscillations to occur in more or less periodicgroups will be shown in many subsequent records.

129

I I I I I I I I I I I I ! I

'1JANl)dO 13sNOs

'NI1 AIUN JASNns,

--

--- - q-s--

=s-S-

_ 7

,==_f

= _

_, . ..._=

-

Y_ .:-<=t

_o:-o=,o

-

5-

0 ~

-z

5-

- ¢

130

Day transmission over short distances, as we have seen fromthe simultaneous record of Figure 6, does not show any markeddifferences when recorded at separated points, perhaps becausethe daytime fluctuations are then too weak to be recorded. But

0C)

C-

20

0~uJLLJcc

F-)

Lli

g

i-iw

z0

H

0

VI-)2

CD0cc-LU)

z0

CLJC-)

L-J

LU

I0

LLi

z

LiJz

z0

0

CC)

z

rfcF

112 +>

c:*0

0 c

at night simultaneous records clearly show transmission varia-tions, which are different for even slightly separated receivers.The upper record of Figure 10 is the normal daytime transmissionfrom Boston to Newton Centre, shown merely for comparison, andessentially similar to the records shown in Figure 6. The two

131

lower records are of night transmission from the same station inBoston, to two points in Newton Centre, separated by 550 meters(1,800 feet). These records are exactly superposed, the an-nouncer check points appearing at 8.42 and 8.52 P. M. on bothrecords, and it will be immediately apparent that there is littlecorrespondence between them. So far as my analysis of suchrecords has proceeded, the only points of similarity are in thelonger swings, which are here very poorly shown because of theflattening caused by the small amplitude and relatively longtime scale.

At the left-hand side of Figure 11 are shown five night recordsof transmission from Boston to Newton Centre, ranging fromhighly disturbed to quiet. At the right-hand side of the samefigure are records made from distant stations on the same nights,and within fifteen minutes of the time of the local station records.It will be seen that on the nights when the record from the localstation showed marked fluctuations, the distant stations showeda predominance of short period swings, while on the quiet nightsthe distant stations gave a predominance of longer period varia-tion.

In Figure 12 are shown fo-ur night records of WNAC, Boston,Massachusetts, as received at Scituate, Massachusetts, 33 kilo-meters (21 miles) distant, of which the first 15 kilometers(9 miles) was over sea water, and the remainder over land. Theserecords also show nights that range from quiet to highly disturbed,and differ in no essential respect from those shown in Figure 11,save for a marked increase in amplitude, caused by the increaseddistance. This increase in amplitude has, however, made ap-parent some of the longer periods, which, altho present in Figures10 and 11, are masked by the flattening.

Figure 13, like the preceding figure, is of night transmissionfrom Boston to Scituate, a range of 33 kilometers (21 miles). Itcovers a period of nearly three hours, and is a beautiful exampleof oscillation groups with nearly quiescent periods between.There is a strong resemblance to a seismograph record of adistant and severe earthquake. The change of axis just before8.30 P. M. is perhaps due to a sudden change of adjustmentof the receiving apparatus.

The four records shown in Figure 14 are at a distance of 68kilometers (42 miles) from station WMAF at South Dartmouth,Massachusetts, to Scituate, Massachusetts, the frequency ofWMAF being 845 kilocycles (wave length 355 meters). Thesefour records are fairly representative of the different night con-

132

WSAC leMONX, SS.K --- YOJ C1R3, KiSS., tII =LsLgRS. OCT. 9. 1923 Hi2 E'ERS oMbMizs 9, 1923.

NOTY CErSCYADY, N.Y. - SITON CERTRE. KASS.,SNAC BOSNON, KSS., --- S ENTO EcRm , KiSS., II XITOa ERS, OCT. 11, 1923, 225 E RLOUNERS, OCTOAERt1I 1923.

13*C BOSTO, iSS-- NSTON CESSRI, KSS., 1t ILOUMETERS, OCS. 21, 1923. NOTCHENECTAYlN Y. --551EWT0 OMI SS.. 225 KIfLONTERS,

ASC BOSTON.0 SS.,Kis . OS CTRE,G HKSS., It KILOMETERS. OCT. 20. 1923. NGY SCHENECTADY. 3.Y . 551102 CENTRE, MASS., 223 KW., OCTOBE 20. 1923.

YICO 3155335351. PA..--- EMON CESTRE, MASS., 760 RY,, OCTOBIN 13, 1923.I0AC BOSTON. MASS.,K - N OI CENTRE. KiSS.. 11 KILOIETMS, OCT. 13, 1923.

0 1 Compar INOreS f e an D t B a S t-

FiGTURE 11-Night-ieComparison Records of Local and Distant Broa4casting Stations

72U H 1MJTUE 28, 1923.

1-00 PM

FIGURE 12 WNAC, Boston, Massachusetts Scituate, Massachusetts, 33km. (21 miles), of which the first 15 km. (9 miles) are over sea water. July 29,

1923

ditions over this range. They differ from the preceding shorterdistance records in that the field no longer fluctuates about anearly horizontal axis, and also in that the intensity occasionallyfalls to nearly zero. South Dartmouth, as normally received atScituate in the daytime, is a strong station with very slight fluctua-

133

1-.01 -2^f --)P

Cd

C~

c.)

a)

C

&-

ct

1:34

tion, whereas at night-time not only does the record show an almostcomplete extinction at times, but the sound entirely vanishes inthe monitoring phones, which shows very conclusively that thenight intensity of a distant station may fall far below the day-time intensity.

1'20 P lM. 730

MWF DARTMOUTH, MASS., ---SCITU I, MASS., 68 KiLaEERs, JULY 31, T923.

I.A

TiQ0- .

FIGURE 14 WMAF, Dartmouth, Massachusetts-68 km. (42 miles), July 27,

135

TWO-Scituate, Massachusetts,1923

AUSTII -COMN

DIST CH FORMU

200

)RWJLL N,DA PERC AGE OF RSE

A, FOR 4L UE!2cOF833 COCY$ES.

300 SCF'IGURE 14A

500

In Figure 14-A I have plotted the Austin-Cohen transmissionformula as a percentage of the simple inverse distance formulaagainst distance, for a transmission frequency of 833 kilocycles(360 m.). At 68 kilometers (42 miles), the distance from SouthDartmouth to Scituate, the difference in electric field betweenthe two formulas is only about 16 percent, which would appearin the voltage-squared ordinates of my records as a change ofsome 35 percent. Yet the record shown in Figure 14 showsvariations which are at times enormously greater than this.Similarly, as will be seen from succeeding records, the transmis-

136

I 00*

909

80*

6of

50*

40*

300

20

10* L0 K.M. 100 600 700

'N

N

N

00

sion from Springfield to Newton Centre, 118 kilometers (78miles) apart, which according to the difference between the Austin-Cohen and the inverse distance should show only 25 percentvariation in electric field, or a change in my ordinates of about56 percent, actually show fluctuations from mninute to minuteof tens or hundreds of times. On the other hand, Messrs. Nicholsand Espenschied2A have found that reception fluctuations inoversea transmission tend to lie between the inverse distance andthe Austin-Cohen formula as upper and lower limnits. It wouldappear from this that overland transmission is subject to muchgreater short period fluctuation than transmission over salt water.

FIGURE 15 WBZ, Springfield, Massachusetts-Newton Centre, Massa-chusetts, 118 km. (74 miles), February 10, 1923

s£' 8°o,A 6' /0 15s a o zs Jo a35 4 4 $Sa

FIGURE 16-WBZ, Springfield, Massachusetts Newton Centre, Massa-chusetts, 118 km. (74 miles), April 18, 1923

!A "Radio Extension of the Telephone System to Ships at Se+i," PRO-CEEDINGS OF THF INSTITUTE OF RADIO ENGINEERS, VolumA 11, Number 2,June, 1923.

137

FIGURE 17-WBZ, Springfield, Massachusetts Newton Centre, Massa-chusetts, 118 km. (74 miles), June 16, 1923

7.50t 35 40 45 So 5s 8 0O ,S x s 0

FIGURE 18 WBZ, Springfield, Massachusetts-Newton Centre, Massa-chusetts, 118 km. (74 miles), September 11, 1923

The four records shown in Figures 15, 16, 17, and 18 are rep-resentative of night transmission from Springfield to NewtonCentre, a distance of 118 kilometers (74 miles). The first tworecords are at 750 kilocycles (400 meters), while Figures 17 and 18are at 890 kilocycles (337 meters), as they are subsequent to thechange in frequencies effective on May 15, 1923. These fourrecords run from quiet to highly disturbed nights, and it is inter-esting in connection with the change in frequency of this stationthat examination of my numerous records before and after thedate of the change does not show any difference which could beattributed to this change.

Figures 19, 20, and 21 are typical of Schenectady to NewtonCentre transmission at night, a distance of 225 kilometers (141

138

FIGURE 19-WGY, Schenectady, New York-Newton Centre, Massachusetts,225 km. (141 miles), February 16, 1923

FIGURE 21-WGY, Schenectady, New York-Newton Centre, Massachusetts,225 km. (141 miles), September 27, 1923

139

miles). We have now apparently reached the transmission dis-tance at which the shorter period fluctuations are at their maxi-mum, altho perhaps this may really lie somewhere betweenSpringfield at 118 kilometers (74 miles) and Schenectady at 225kilometers (141 miles). From now on the records shown will havea predominance of the longer period swings.

Figures 22 and 23 are representative of night transmissionfrom WJZ, New York City, to Newton Centre, a distance of 290kilometers (181 miles). Figure 22 is of the opening night of Broad-cast Central (as duplex stations WJY and WJZ are termed), andthe interest taken in this by mny Newton neighbors is shown bythe "squeal" record appearing just above the time scale markingsat the lower edge of the sheet. Each tinme a beat note from aneighboring oscillating receiver was heard in the monitoringtelephones, a spot of light was flashed on the record sheet. Thesesqueals are noticeably grouped most closely around low spots inthe transmission, and also around any interval when the stationwas not modulating. It is a fixed idea of the average broadcastlistener that when signals are weak or absent his receiver must besuspected and adjusted. At this distance of transmission thecharacter of the record begins to change, and the longer periodswings, measured in minutes, begin to become more prominent.

2,5' h@ J , ,, /O ,15FIGURE 22-WTJZ, New York City Newton Centre, Massaclhusetts, 29( km.

(181 miles), May 15, 1993

Figures 24 and 25 are KDKA, Pittsburgh, Pennsylvania, asreceived by night at Newton Centre, a distance of 760 kilometers(475 miles). The distance is now great enough to bring outstrongly the long period swings, of about five and a half minutesin Figure 24, and about ten minutes in Figure 25.

140

On quiet evenings, as will be seen from later records of thesame station, the long periods are even more in evidence.

This record is typical of night transmission from WCAP,Washington, District of Columbia, to Newton Centre, a distanceof 630 kilometers (394 miles). Altho the distance is less than that

8 206. 2 30 31 8i 4S 5- 5s 9o so's-

FIGURE 23-WJZ, New York City-Newton Centre, Massachusetts, 290 km.(181 miles), September 16, 1923

of the records shown in Figures 24 and 25, the transmission fromthis station is always relatively free of fluctuations having aperiod of one minute or less. It would seem that the groundroute followed by the transmission played some part in deter-mining the character of the variations.

FIGURE 24-KDKA, Pittsburgh, Pennsylvania-Newton Centre, Massa-chusetts, 760 km. (475 miles), February, 14, 1923

141

I II $ IIIIIIIII111 IIIIIIIIII I IIIHI III) ) H I H HIIII)4S5 So *-s 9500 s )o is 20 .15 J?o .1v'

F'IGUTRE 25-KDKA, Pittsburgh, Pennsylvania Newton Centre, Massa-chusetts, 760 km. (475 miles), March 22, 1923

FIGURE 26 WCAP, Washington, District of Columbia-Newton Centre,Massachusetts, 630 km. (394 miles), September 1, 1923

Figure 27 is of night transmission from WDAP, ChicagoIllinois, to Newton Centre, Massachusetts, a distance of 1,400kilometers (875 miles). A long swing of about twenty minutesis clearly brought out in this record, with superimposed periodsof about thirty and sixty seconds. This record was made on amoderately disturbed night; on a quiet night the curve is muchsmoother.

ic/op0 15 /Q 2S so 40 45 sa 3d 1100

FIGURE 27-WDAP, Chicago, Illinois-Newton Centre, Massachusetts,1,400 km. (875 miles), August 2, 1923

142

Figure 28 is of night transmission from WBAP, FortWorth, Texas, to Newton Centre, Massachusetts, a distance of2,600 kilometers (1,630 miles). A long period of some fifteenminutes is prominent, and with the rather high amplificationnecessary for this distant station, the background disturbancebecomes quite appreciable.

1-tS"b. l J oQ 4/- *a0 s 02.00 M 1 0

FIGURE 28-WBAP, Fort Wortb, Texas-Newton Centre, Massachusetts,2,600 km. (1,620 miles), February 8, 1923

Figure 29 is a simultaneous night record at Newton Centreof Schenectady, distant 225 kilometers (141 miles), taken atreceiving points separated 550 meters (1,800 feet). With theexception of the period of high intensity at about 8.10 P. M., itis difficult to identify any of the detail of one record with that ofthe other. The longer period swings of these records are, how-ever, substantially the same, and this becomes apparent if thecurves are smoothed.

The fact that altho the shorter period swings are not thesame on records made at slightly separated points, the longerperiod fluctuations are substantially identical, is well broughtout by Figure 30, which is of night transmission from Pittsburgh,Pennsylvania to Newton. This was made on a quiet evening,so that the short period variations are weak, and altho consid-erable distortion occurs, particularly around 8.50 P. M., the mainfeatures of the two records are the same.

If, however, the receiving points are at all widely separated,there is little or no similarity between their records of a distantstation. In Figure 31 the transmitting station is WBZ, Spring-field, Massachusetts, and the two receiving points are NewtonCentre and Scituate, Massachusetts, 118 and 155 kilomneters(74 and 97 miles), respectively, from Springfield, and 41 kilometers(26 miles) from each other. Aside from the general increase inintensity after sunset, which was at 7.20 P. M., there is no cor-respondence between these two records.

143

- z

CQ

4-~~

== _0

0I

5j

1i4

cmCC\e

V)L--

C-j

N0

(I-)

4:iL,JcczL-Q-)

z0

LL

Coio

z

04:

f)F-D

C-

IIZI-

r--

/Sl t~~~i

\hW - J~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~I

850 PMKDKA PITTSWRGH, PA - -- NEWTON CENTRE, MA55 7C0KK PICKARD

FIGURE 30 KDKA, Pittsburgh, Pennsylvania--Newton Centre, Massa-chusetts (760 km. (475 miles), SHAW

/\)p

FIGURE 31 Simultaneous Reception of WBZ, Springfield, Massachusetts,at Newton Centre, Massachusetts, and Scituate, Massachusetts, Distant118 and 155 km. (74 and 97 miles). Scituate is 41 km. (26 miles) from Newton

("entre. June 9, 1923145

o

Figure 32 is a simultaneous night record at the same receivingpoint in Newton Centre, of two stations in Newark, New Jersey,distant 315 kilometers (197 miles), and, of course, operating atdifferent frequencies. There are no points of correspondencebetween these records, save accidental coincidence of maximaand minima which would occur in the superposition of any tworandom complex curves.

82O. 253i 3S 40 4S sa 5s 9500 5

FIGURE 32 Simultaneous Record at Newton Centre, Massachusetts, of WORand WJZ (original location), both at Newark New Jersey, Distant 315 km.

(197 miles), March 2, 1923

Figures 33, 34, and 35 illustrate simultaneous transmissionto Newton Centre from widely separated broadcasting stations.As in Figure 32, there is a complete lack of correspondence.

3933a . 35- 40 S4o So s, / 5'o/s I zo

FIGURE 33-Simultaneous Record at Newton Centre, AMassachusetts, ofWEAF, New York City, and WJZ, Newark, New Jersey, Distant 290 anDd

315 km. (181 and 197 miles), March 10, 1923

146

8 tot 2r Jo 35 'O 45- so sr 9.oo s 0lo Is

FIGURE 34-Simultaneous Record at Newton Centre, Massachusetts, of WBZ,Springfield, Massachusetts, and WEAF, New York City, Distant 118 and

290 km. (74 and 181 miles), February 24, 1923

,;, 4 S 1~~S,/ '\, tsWI4L493 AU4 +V4j -1 H1- - tH- HIIIJI KIII II-L I.aPS 9 JS Io 45 S' rSS 00Xo /O 15

FIGURE 35 Simultaneous Record at Newton Centre, Massachusetts, ofCKAC, Montreal, Canada, and WEAF, New York City, Distant 405 and

290 km. (254 and 181 miles), February 24, 1923

In Figure 36, which is a simultaneous night record at NewtonCentre from New York City and Springfield, there is one cor-respondence; namely, the dead period beginning about 8.42 P. M.Unfortunately for further possible points of similarity, Spring-field signed off at 9.04 P. M.

I have said above that reception from a distant station is notthe same at separated points, unless the separation is very smallindeed. In addition to extremely small separation, it is alsonecessary for identical records that the two antennas be of thesame directional properties, and similarly oriented.

In Figure 36-A is shown a simultaneous record on two openantennas, separated only 30 meters (97 feet) on centers. The

147

two records are substantially identical, and the slight differencesbetween them are probably due to the fact that one of the openantennas was a large T, while the other was of the inverted Ltype, so that there was a slight difference in directional properties.

ZS itoh 35* 4 5O3S0 5ss- 3oe sf 'a

FIGURE 36 Simultaneous Record at Newton Centre, Massachusetts, ofWEAF, New York City, and WBZ, Springfield, Massachusetts, Distant 2990

and 118 km. (181 and 74 miles), March 10, 1923

FIGURE 36A WNAC, Boston, Massachusetts, receivedat Newton Centre, Massachusetts, 11 kilometers (7miles) distant, November 13, 1923, on antennas 30 meters(97 feet) apart on centers. Upper record on large openantenna by Shaw; lower record on small open antenna by

Pickard. Record is 4 minutes and 40 seconds long

FIGURE 36B-WNAC, Boston, Massachusetts, received at NewtonCentre, Massachusetts, 11 kilometers (7 miles) distant, on Novem-ber 13, 1923, on two small loops, 3 meters (9.7 feet) apart on cen-ters and in line with WNAC. Upper record by Pickard; lower record

by Shaw. Record is 6 minuites, 20 seconds

Figure 36-B shows a simultaneous record on two small loops,3 meters (9.7 feet) apart on centers, and each loop in line withthe broadcasting station. Here, also, the records are substan-tially identical, the slight differences which appear being probablydue to differences in the open antenna effect of the two loops,which were compensated.

148

But in Figure 36-C, which is a simultaneous record made onantennas of different directional properties, the upper recordthat of a T open antenna, and the lower on a small loop, separated15 meters (48 feet) on centers, a total dissimilarity appears, bothin shape and amplitude, the loop record showing about fourtimes the amnplitude of the open record.

FIGURE 36C WNAC, Boston, Massachusetts, receivedat Newton Centre, Massachusetts, 11 kikometers (7 miles)distant, November 13, 1923, on antennas 15 meters (48feet) apart. Upper record on open antenna by Pickard;lower record on loop by Shaw. Record is 5 minutes long

Figure 36-D is a double record made on two small loops,uncompensated, 3 meters (9.7 feet) apart on centers, at rightangles with each other and each loop at 450 from the plane, in-cluding the broadcasting station. Here, also, the two recordsare entirely dissimilar.

FIGURE 36D-WNAC, Boston, Massachusetts, received at Newton Centre,Massachusetts, 11 kilometers (7 miles) distant, November 13, 1923, on twosmall loops, 3 meters (9.7 feet) apart on centers. These two loops were atright angles with each other, and at 450 with the radio bearipg of the BostonStation. The upper record was made by Pickard on a loop in a NE-SW plane;the lower record is by Shaw on another loop in a NWX-SE plane. Record is 9

minutes long

It has been thought that reception variations in loop recep-tion were largely due to changes in the direction of wave-front,and this may be true in some localities, and with certain distantstations. For example, I have often found large short-periodvariations in the apparent bearing of ship stations at night, whenthe receiving point was near the water's edge, and the line of

149

transmission ran grazingly along the shore. Under such con-ditions an apparently weakened or lost signal could sometimesbe restored by swinging the loop thru 90°. But inland, and par-ticularly at my principal receiving point in Newton Centre, over-land reception from broadcasting stations does not show largechanges in wave-front. A loop null point taken on WNAC,Boston, Massachusetts, 11 kilometers (7 miles) distant, remainsconstant thru an entire evening to within a degree or two, sothat the effects shown in the preceding figures can hardly be dueto changes in wave-front.

Altho the relation of these short period flucuations to trans-mission frequency is a matter which I must reserve for a laterpaper, I have shown in Figure 36E an interesting simultaneousrecord, made at Newton Centre on two open antennas separated30 meters (97 feet) on centers, of the two transmission frequenciesKDKA, Pittsburgh, Pennsylvania. The upper record is at 920

9100 PM. 910o

FIGURE 36E-KDKA, Pittsburgh, Pennsylvania, received at Newton Centre,Massachusetts, 760 kilometers (565 miles) November 28, 1923. IUpper record

920 kilocycles (326 miles); lower record 3,200 kilocycles (94 miles)

kilocycles (326 meters), while the lower one is at 3,200 kilocycles(94 meters). I selected this particular record because it was outof the ordinary in that there is a general resemblance between thecurves; ususlly reception at 3,200 kilocycles is subject to a muchmore rapid variation than is here shown. Figure 36F, whichis a four-hour record of KDKA at 3,200 kilocycles, is more nearlyrepresentative of this transmission.

150

FIGURE 36F-KDKA, Pittsburgh, Pennsylvania, received at Newton Centre,Massachusetts, 760 kilometers (565 miles), 3,200 kilocycles (94 meters),

November 29-30, 1923

I have devoted some little study to KDKA's 3,200-kilocycletransmission, and the result, at least in so far as Newton Centrereception is concerned, may be briefly summed up as follows:In the afternoon and early evening reception from KDKA isoften better at 3,200 kilocycles than at 920, while later in theevening the lower frequency is usually better. Until recentlythe higher frequency was quite free of disturbances; now there isa noticeable amount of amateur working at or near this frequency,and a sufficient number of oscillating receivers to much thequality badly at times.

It has probably been observed by all broadcast listeners thatin general the quality of reception from a distant station is mark-edly inferior to that obtained from a local station, this inferiorquality being a matter entirely aside from impairmant of recep-tion due to disturbances. I am now convinced that the principalcause of the distortion in reception from distant stations is thatthe short period variations in transmission are not the same forthe carrier wave and its two side bands, and I hope to presentin another paper certain records which show this effect. For-tunately there are two obvious remedies for this form of distor-tion; the use of a single side band for transmission, and the em-ployment of much higher transmission frequencies. Neither

151

of these remedies will be immediately popular with the averagebroadcast listener.

So far as the varying intensity of the sound in the telephonereceiver is concerned, there are several simple expedients whichwill markedly smooth out reception from a distant station.Thus, the grids in a radio frequency amplifier train may be floatedon a fairly large condenser, shunted by a high resistance. Whenreception is weak, the grids assume a small negative potential,and amplification is at a maximum. When the input rises, aa large negative charge is built up of the grids, and amplificationis reduced. Similarly, a separate rectifier connected to the out-put end of the radio frequency amplifier and to the grids will havethe same effect.

So far I have confined myself strictly to my records, whichhave shown you graphically the facts of short period variationsin broadcast transmission. There is as yet no adequate theoryas to their cause. But I am sure you will agree with me as to theutility of hypotheses. Scientific progress is literally milestonedwith hypotheses, most of which are to-day gravestones. But anhypothesis is at least something concrete, and if it stands theattack of new facts, and the further analysis and correlation ofour present knowledge, it iuay form a stepping-stone to thetruth. I have already iuentioned a further paper on this subject,and in this I hope to present, among other things, the curiousresults of harmonic analysis of these record curves, some at-tempted correlation between these fluctuations and other ter-restrial happenings, and simultaneous records of field intensityat a large number of separated points. I also hope to establishrather definitely the size and shape of the area within which thesevariations are identical. But at the present time I feel verystrongly that any attempt at a detailed hypothesis would bedecidedly premature, altho some rather general conclusions maynot be out of place.

A very odd explanation of these variations has been gainingmuch vogue in the popular scientific press, to the effect that theyare due to other receivers in the neighborhood, particularly ifthese happen to be of the much maligned single circuit regenera-tive variety. I have many records of transmission, principallymade by audibility meter measurement, which run back overfifteen years, when the density of receivers was far less than atpresent. These records show exactly the same short period fluc-tuations that now exist, so that this explanation does not seemvery plausible. However, I have recently made a number of

152

records of reception from distant stations under conditions ofsevere exposure to nearby regenerative receivers, of which Figure37 is typical.

111111s1 11 llilrrrllgll~I 1T]/3'i 40 f4's~ Sa~ Sr 0.00 FF5 F1° lS

FIGURE 37-WBZ, Springfield, Massachusetts, Received at Newton Centre,Massachusetts, Distant 118 km. (74 miles). Base lines as at 7.37 to 7.42 P.M.Indicate Periods of 5 Minutes each when Single Circuit Regenerative Receiver

in Same House and with Adjacent Antenna was in Operation. September 10,1923

In this record, the broken heavy base line indicates periodswhen a single circuit regenerative receiver in the same house withmy recording set, and with a fair-sized open antenna adjacentto the one employed with the recording set, was in operation.Full regeneration was used almost to the point of oscillation,but this record, like several others which I have made with dif-ferent exposures of antennas, shows no effect whatsoever.

The fluctuations in radio transmission, at least those of shortperiod, are at first thought most readily explained by the as-sumption of plural transmission paths; that is, the effect appearsto be one of interference. I believe that De Forest was the firstto suggest this in 1913,3 for short period variations in night-time transmission from the arc stations of the Federal TelegraphCompany on the Pacific coast. Such plural paths may lie inaltitude or azimuth, or both, but the favorite explanation hasbeen that reflection takes place at some high level in the atmos-phere, because of an ionized stratum called "the Heaviside layer."

In order to deflect electromagnetic waves, we may reflectthem from a conducting surface or mirror, refract them by achange in optical density of the medium, or bend them by a two-of three-dimensional periodic structure of absorbing and non-absorbing elements. Inasmuch as, in our atmosphere, all threeof these methods call for ionization, and as the maintenance ofan ionized state requires energy, it is easy to arrange thesemethods in decreasing order of energy required to produce theobserved effects, which would also seem to be the order of increas-ing probability.

3 PROCEEDINGS OF THE INSTITUTE OF RADIO ENGINEERS, volume 1, num-ber 1, 1913, pages 42-51.

153

Maxwell's theory for the relation between the reflectingpower of a mirror and its electrical conductivity4 requires for theeffective reflection of the waves employed in broadcasting a con-ductivity somewhere between fresh water and sea water. Eccleshas computed5 the conductivity necessary for refraction of suchwaves to be out of the order of one-millionth that of sea water.I have approximately calculated the ionization required to reflectthe waves on the assumption of a periodic structure; a sort of"mackerel sky"* of ionized clouds, with the result that a con-siderably smaller number of ions is required than for refractionby a uniform state of ionization. Because of the smaller amountof energy required to maintain suLch a stratified state, I haveconsidered this the more probable upper-level condition capableof causing reflection-like effects.

Inasmuch as I have found well-defined short period varia-tions of more than ten percent amplitude in transmission overa distance as short as eleven kilometers (6.9 miles), in which thedirect path from transmitter to receiver was at least ten timesshorter than the total "reflected" path, it would seem to involvenearly normal reflection with an efficiency of one hundred percentor over, on the conventional Heaviside layer hypothesis. If somerecent' revisions of our ideas as to the density of our atmosphereat high levels are correct, the Heaviside layer is perhaps twiceas high as we have supposed, which apparently removes it as apossible explanation of fluctuations over such short distances oftransmission.

In view of the fact that night-time transmission over greaterdistances than eleven kilometers (6.9 miles) not infrequently fallsto much lower levels than at any time during the day, it seemsnecessary occasionally to invoke the assistance of some form ofplural path transmission with its consequent interference effects.But the general tendency of night-time field values is to range be-tween the normal daytime transmission value as a lower limit, andthe value given by the transmission formula without the exponen-tial or so-called absorption term as an upper limit. It wouldseem, therefore, that the principal factor affecting radio trans-mission was absorption, with reflection-like effects present,butplaying a minor role.

4Drude, "Physik des Aethers," page 574.5 Eccles, "London Electrician," 69, pages 1015-1016, 1912.*A "mackerel sky" is a dappled cloud formation consisting of heaped-up

or rounded clouds arranged in wisps or streamers-EDITOR.6 Dobson, "The Characteristics of the Atmosphere up to 200 Kilometers

as Indicated by Observations of Meteors," "Quarterly Journal of the RoyalMeteorological Society," volume XLIX, 207, July, 1923.

154

It may at first seem altogether unlikely that such rapid fluc-tuations as I havefound in transmission could be caused by changesin absorption. Absorption can only be explained as due to ioniza-tion, and while it does not seem impossible that ionizationchanges large enough to affect our whole atmosphere might havea diurnal period and so cause the well-marked difference betweenday and night transmission, it is difficult to imagine such largescale effects varying with periods of minutes and even seconds.One difficulty in the past has perhaps been our tendency to at-tribute too much of the ionization to the effect of direct sunlight,and too little to the effect of alpha particles shot out by the sun,which continually fall in drifting clouds into our atmosphere.These charged particles may arrive intermittently and at shortintervals, perhaps in groups of small, cloud-like masses, and be-fore the outer limit of our atmosphere is reached the earth'smagnetic field begins to deflect and comb them out along its owndirection. This process continues, perhaps, down to an elevationof some 150 kilometers (94 miles) or less, eventually drawing theseclouds out into long streamers in a south to north direction.Then, perhaps, these streamers are captured by the high levelair currents, which over the United States at heights of between100 and 20 kilometers (63 and 13 miles) drift slowly from westto east, and, finally, they become subject to the stronger and vary-ing low level winds.

I have attempted to illustrate this idea in Figure 38. FromA to B the clouds of charged particles are drawn out in a south-to-north direction as they fall; from B to C they are deflected atnearly right angles to this course, or in a west-to-east direction,and from C to D thay are at the mercy of the variable lower levelwinds. This results in the development of two striated layers,the upper one running south-to-north, the lower one west-to-east. Some real vision of the upper level structure isperhaps given to us by the aurora, which may be simplyelectrical discharges following and lighting up these ionizedpaths.

Such periodic structures as I have outlined could act asgigantic gratings, capable of reflecting as spectra the radiationfrom distant transmitters, and thus causing interference effects.For stations to our west or east, the upper layer would be moreeffective; for southern or northern stations the lower, because ofthe direction of the grating elements. According to this, thereshould be some difference between east and west transmissionon the one hand, and north and south on the other. A com-

155

*::9

*:.55

555* 55

D1 :@*le

55,

SSS

*:* \S

FIGUR1E 38

parison of a typical west-to-east transmission, such as that ofFigure 27, with a typical south-to-north transmission does in-deed indicate a difference in the character of the variations.

It is believed that at least a portion of the earth's magneticfield is due to these descending charges, and if this is so, we shouldexpect to find some correspondence between magnetographrecords and these transmission records. A magnetograph recorddoes in fact show seasonal and diurnal changes, disturbancesof great amplitude at times of intense aurora, and it also showsa variety of short period pulsations, measured in hours, minutes,and even seconds. Periods of 2 to 4 minutes are most frequent,while pulsations following each other at about 30 second inter-vals are not uncommon. Some days are relatively quiet, that is,free from rapid large amplit-ude oscillations, others are moder-ately disturbed, and many are highly disturbed. All of theseeffects are known in radio transinission, and somne of them areshown in my records. In a later paper I hope to present simul-taneous transmission and magnetograph records.

My present explanation of the transmnission changes presented156

in this paper is that they are due primarily to changes in absorp-tion, and secondarily to reflection-like effects.

Both absorption and reflection are caused by ions in ouratmosphere, which I assume are not uniformly distributed, butarranged in streamer-like clouds, oriented in the upper levelsby the earth's magnetic field, and in the lower atmosphere byair currents. During the day, the direct radiation from the sunadds to this ionization, producing thereby a sufficient uniformityin its distribution so that it acts principally to absorb, therebymasking any reflection effects which would be apparent in itsabsence. Before sunset the solar ionization decreases, and atnight this additional source of ionization vanishes, and the de-scending ionized structure is unmasked, varying from minute tominute in its effect over a given locality as it drifts with the lowlevel air currents. At times this structure becomes sufficientlyperiodic to form a two- or three-dimensional grating, and theninterference effects are found, including not only transmissionminima lower than any daytime value, but also changes in wave-front. Over short distances of transmission, individual or smallgroups of descending ions at very low levels may account forthe short period swings which predominate in such transmission,so that the effect may be confined almost entirely to the loweratmosphere. Over greater distances, the effect would be a sortof statistical average of great numbers of these ion clouds, whichwould tend to smooth out the shorter period effects, and accen-tuate the longer periods corresponding with the arrival of largegroups over large areas. Certainly there should be a markedcorrelation between transmission and weather.

Nipher7 has found a well-defined correspondence between thehorizontal intensity of the earth's magnetism and very localatmospheric disturbances. In Figure 39 his magnetographrecords between 4 and 5 P. M. a sharp dip coinciding with thepassage of a dense cloud. Other records show the effect of localair movement and small aread showers. Nipher's conclusion isthat "local variations in the earth's magnetic field are determinedwholly by local weather conditions"-perhaps the apparentlyvery local short period variations in radio reception may becaused by changes in a comparatively minute volume of low levelatmosphere.

In conclusion, I wish to acknowledge my indebtedness to Mr.H. S. Shaw, Jr., for his invaluable co-operation in this work, not

I Nipher, "Variations in the Earth's Magnetic Field," "Trans. Acad. ofSci. of St. Louis," volume XXII, number 4, 1913.

157

FIGURE 39

only in the design of portable apparatus for recording and in theactual taking of many of the records, but for his many helpfulsuggestions.

SUMMARY: Continuous records of electric field intensity at distances offrom 11 to 2,600 kilometers (6.9 to 1,650 miles) from various broadcastingstations are made by amplifying and rectifying the received current from anopen antenna, and recording galvanometer deflections produced by this recti-fied current. These records show a complex curve containing periodicitiesranging from seconds to tens of minutes, the relative amplitude of the shortand long period elements varying with the distance. The transition from dayto night conditions is shown, and well-marked short period variations ("fad-ing") in night-time reception from a station only 11 kilometers (6.9 miles)distant are found. Simultaneous records at separated receiving points arefound to be dissimilar, even with separations as small as 550 meters (1,680feet), and simultaneous records of different distant stations taken at the samereceiving point are also shown to be entirely unlike. A tentative hypothesisfor these fluctuations is advanced by the author, in which verying local absorp-tion plays the principal part, but with marked interference effects producedby plural path transmission.

158