Paris (France). Bureau des longitudes.

Wireless time signals. Radio-telegraphic time and weather signals transmitted from the Eiffel Tower, and their reception online

. (page 3 of 8)
Online LibraryParis (France). Bureau des longitudesWireless time signals. Radio-telegraphic time and weather signals transmitted from the Eiffel Tower, and their reception → online text (page 3 of 8)
Font size
QR-code for this ebook

if the instrument beats the second, and a dot, for
instance, by the mark s, equal in size to about one-

* The decisions of the Conference Internationale de I'Heure can only
come into force after being approved by the governments represented
at this Conference. Details of these scientific signals are given in
Chapter III.


quarter the interval between the numbered strokes.
To obtain the correction to be made to the time re-
corded by his instrument, the observer first of all must
note the number of the second which immediately
precedes s, say i second, and estimate the interval
of time which separates this second from s, say 0*2
second. He must then read the minute and the
hour corresponding to the signal s, say loh. 44m.
If the signal J is that of loh. 4Sm. , the correction
for the watch or clock will be

(loh, 45m.) — (loh. 44m. i*2s.)= +58-8 seconds.

In other words, he must add 58 "8 seconds to the
time recorded by his instrument. The computation
of this fraction of a second cannot be made with
great accuracy, for there is the difficulty of mentally
isolating the commencement of the dot, then of plac-
ing it with reference to the seconds' beats which
include it, with only the rhythmic succession of
beats as a guide.

It is easy enough, however, even for anyone
with little practice, to estimate it to at least half a
second ; for an expert and somewhat gifted observer
the error is rarely more than 0"2 second. To
place the beginning or end of a dash proceed in a
similar fashion.

Corrections. — To get the total error, it is ad-
visable to add the error of the time signal itself to
the error of estimation just indicated. To calculate
its average value, the various causes which may


influence it must be examined. The time calculated
by means of the rate of the time-keeping clocks, and
by which the signal-sending clock is regulated, is
affected by an error as great as the number of clocks
used in the extrapolation is small, and their quality
mediocre, and generally this error increases with
the time which elapses after the last adjustment.
When this time exceeds fifteen days, as is not un-
common in Paris in the winter months, the error
may be as much as i second. To reduce it to a
minimum, there is only one thing to be done, and
that is to diminish the time of extrapolation as
much as possible by utilising the readings obtained
in other observatories. It will be seen later how
the use of scientific time signals will enable several
observatories to co-operate without difficulty in
determining the time. Under these conditions
the time of extrapolation of the rate of the time-
keeping clocks will never be more than a few days,
and the error of the time calculated will always be
very small, O'l second at most.

With the present arrangement (fig. 17) for
sending time signals from the Eiffel Tower, in order
to regulate the clock sending the signals according
to this extrapolee time, the following must be
taken into account : —

I. The difference between the moment when the
clock beats the given time for sending the
time signal, and the time when the electric
contact W is closed (fig. 17).


2. The delay, with reference to the closing of

the contact W, and the production of the
spark in the spark-gap G, i.e. of the
emission of Hertzian waves.

3. The time which elapses between the moment

when the waves are launched into space,

and that when the corresponding sound is

perceived by the observer in his telephone


When the despatch of the signals is made

automatically, notice must also be taken of the

error caused by the automatic transmitter, the

contacts of which may close either too soon or too

late with respect to the clock. This error must be

periodically determined. The difference between

the moment of closing of the contact W and the

corresponding beat of the clock U is apparently not

determinable with much accuracy, owing to the

arrangement of the system which produces the

contact ; it may be obtained approximately by

measuring the duration of the contact with the

end of which the beat appreciably coincides.

The retardation in the production of the spark
with respect to the closing of the contact W or of
the automatic transmitter contact is due to the
mechanical and electrical inertia of the intervening
instruments. It consists of the retardation in the
attraction of the relay plate, the retardation in the
closing of the key M, and the time necessary for
the current of the feeding circuit to produce at the


armatures of the condenser K a potential difference
sufficient to cause the spark pass at G.

Finally, the retardation in the perception of the
waves in the telephone is analysed in the following
manner : (i)Time necessary for the propagation of
waves between the points of emission and reception
(negligible in practice) ; (2) interval separating the
moment the waves strike the aerial from the
moment the corresponding sound is heard in the
telephone by the observer. There is no occasion to
carry this analysis further, the process hereafter
given permits of the easy measurement of the total
retardation from the moment the contact W of the
clock transmitting the signals is closed, up to that
when the sound is perceived in the telephone

This process is based on the use of the method of
coincidences which is already utilised for the accurate
comparison of two time-measuring instruments.

Principle of the Method of Coincidences. — The
following is the principle of this method: — Let A
and B be the two time-measuring instruments to
be compared. Draw an indefinite straight line Ot
to represent the axis of time, starting from an origin
O corresponding to the moment taken for the initial
period. On this line mark off lengths proportional
to the times of the successive beats of A and B.
The diagram shown in fig. 21 is obtained. The
numbered marks above O^ represent the beats of
A, those below it of B. The intervals between


the beats of A are supposed to be equal to each
other, as also are those of B, the latter being shorter
than the former by -^ of a second. By listening
either directly or by the intermediary of telephones
to the collective beats of the two series, the beat
of B which follows that of A is heard to approach
it gradually, then almost coincide with it, pass it,
and gradually diverge from it more and more.
Note the times A^ and k^ of the" beats of A and B
which most nearly coincide. If the period e which
separates these is small, neglect it and take as a
comparison h^—h-^ either for the time h^^ of A or

&^ 6 T e S 10 II 12 !3 t^ IS 16 17 Ift 19

^°i ' I' I ' I ' I ' I ' I I. 'i — \—\ \ — H — H — r^— rr

•f" 17 IB 19 20 Zl 22 2S V, 2S 26 tl 2t 29 30 31 il

Fig. 21.

for that of ^ of B. Graphically, the method of
procedure is as follows : — Trace the beats of A,
commencing with, say. No. 8, up to the one which
coincides best with a beat of B ; No. i6 of A is
seen to follow No. 28 of B very closely, whilst 15,
which came before 27 of B, is more markedly
separated from it ; and for comparison with the time
16 beats of A or 28 beats of B (16-28), beats are
taken, neglecting the hours and minutes. The
operation, as will be seen, is the same as reading a
vernier, only, as there is no obligation to take a
fixed beat of B as the zero of the vernier, by
preference that of the coincidence is taken. The
approximation obtained by this method is calculated


exactly like that of a vernier. If this comprises
n divisions equal exactly to (n—\) divisions of the

scale, each of them is equal to ( i — j scale divisions,

and the error in reading is at most equal to half
the excess of a scale division over a vernier division

or — scale division. Graphically, the maximum

error in comparison by coincidence is at ^^ beat

of A.

In a general way, if the interval of the beats of

B, which most nearly approaches the interval T^ of

the beats of A, has a value T^^^idz-j, n being any

number whatever, the separation e of the two

beats which most nearly coincide is at most equal

to ^.


In practice the accuracy attainable by the method
of coincidences is as rapidly limited as that of the
vernier. When n is rather large the observation
of the coincidence becomes difficult and in part
illusory, if the sound of the beats has a duration
which is not negligible. This duration is graphi-
cally represented in a greater or less thickness ol
the marks which represent the beats. As a result,
several consecutive divisions of A and B apparently
coincide. This is the principal but not the only
cause which affects the accuracy of the estimation
of the coincidence. If the beats of one series are



louder than those of the other, the latter are
obscured in the region of the coincidence. The
sonorousness of the beats, their difference of pitch
and tone are equally unfavourable for the close
estimation of coincidences.

Determination of the Retardation of a Time
Signal with regard to the Corresponding Contact
of the Signalling Clock. — To apply this method
to the determination of the retardation of the

ZiHfsWMn ^}Ml)iAir

Fig. 22.

signal perceived in the telephone with relation to the
closing of the contact W, the operation is as follows :
— For the clock U, or for the automatic transmitter
which does not lend itself to the transmission of
rhythmic signals, substitute an auxiliary clock U^
(fig. 22) whose period can be regulated. This is
provided with an electrical contact Wj^, similar to
that W of the first clock, but closing at each beat for
a short and adjustable period. The other portions
of the apparatus require no modification. By
conveniently adjusting the duration of the contact,


a wireless transmission spark is obtained at each
complete oscillation of the new clock. This series
of unique sparks is heard in the telephone T of a
wireless receiver V placed close to the clocks.
A second auxiliary clock U^, like the preceding one,
but regulated to give a period differing from it by
w to Y^^ of a second, is likewise fitted up. The
contact Wg of U^ is connected with a battery /^j a
condenser Kg (telephone type) of 3 mfds. , shunted
by a resistance r^ of 30,000 ohms, and finally to
the primary b^ of a small telephonic induction
coil which is movable inside the secondary. This
latter is put in series with the secondary of a
similar coil b^ and both secondaries joined up 'to
the terminals of the telephone T of the wireless
receiver V above mentioned.

Each time the contact Wg closes, the condenser
Kj is suddenly charged ; the current from this
charge generates an induced current in the secondary
of b^ and produces in turn a sound in the telephone
T. When Wj is opened, no sound is heard in the
telephone. The condenser being discharged through
the resistance r^ again becomes charged with
another closing of the contact, and again a sound
is perceived. Each complete oscillation of Uj is
therefore distinguished by a clear and unique sound
in the telephone T. Considering the electrical
elements in the circuit, the interval which separates
the moment of perception of a sound in the telephone
from that of closing the contact is negligible.


Finally, a circuit similar to the above is established
in connection with the induction coil b.^. A double
commutator N enables contact Wj of U^ to be con-
nected either to this second circuit or to the wireless
transmitting apparatus.- All the apparatus being
thus arranged, commence by placing commutator N
across the left-hand contacts. The clock U^ produces
a series of radio-telegraphic dots which, acting on
the receiving aerial, are perceived in the telephone
T at the same time as the beats of Ug. Move the
primary b^ more or less into the secondary so as
to equalise the intensity of these latter beats with
the radio-telegraphic beats of Uj and observe their
coincidences. Take a suiiScient number of them
to obtain, with as much accuracy as desired, their
time a in time of U^ for instance, and immediately
after having written down the time h of the clock
Uj of the last, move the commutator to the right-
hand contacts. The closings of the contact Wj of the
clock Uj then produce in the telephone C through
the intermediary of b.^ beats similar to those of Uj.
Their period is evidently equal to that of the radio-
telegraphic beats previously observed ; the interval
of their coincidences with the beats of Ug is there-
fore still a. But the first coincidence, instead of
happening at a time h-\-a of Uj as would have
been the case if the commutator N had not been
moved, now happens at a different time Ji . This
time H is less than h-\-a, that is, that the coincidence
comes earlier if Uj beats more quickly than U^ ;



on the other hand, it happens later if Uj beats more
slowly than U^. To make this clear, let us suppose
that the period T^^ of Uj is shorter than that T^ of
Ug, and state

T, = T,(i-i)

The gain A + a — Zi' evidently represents the
number of times the retardation required (p) con-
tains Tg — Tj and we get

n— I
If we have

the retardation p is expressed by

p = {k' -h-a){T^-T^) = {h' -h-a)^

This test has been carried out at the Eiffel Tower
and enables us to state that with the key used at
present the retardation is from yf ^ to -^^ second.

It is easy to take this into account when setting
the signalling clock to time by putting it forward
a fraction of a second equal to p.

To sum up, the preceding considerations show
that the errors which affect the time obtained by
means of a time signal are reduced to the two
following : — The error in the transmitted signal
coming from the extrapolation in the rate of the
clock, and the error in estimating the fraction of


a second between the full second which precedes
the signal and the signal itself, which error is made
by the observer. Except in quite special cases,
this latter error is rarely less than ^^ second.
In general, the exactness of the legal time obtained
by transmission of time signals cannot be reckoned
to nearer than -^ second : that is the limit for an
expert observer.

For an inexperienced observer this limit becomes
about J second at least.

Transferring the Record of a Printing Chrono-
graph, such as is obtained by Astronomical Obser-
vations, to that of a Time-keeping Clock. — Previous
remarks have been based on the assumption that
astronomical observations directly supplied the
reading of one of the time-keeping clocks ; that
is to say, the correction of the time recorded
by the clock at the moment of any beat, ordinary
or electrical, if the clock is provided with con-
tacts. This is what takes place, ignoring the
personal equation of the observer when the obser-
vations are made by the method of the eye and
the ear. It is not nearly the same when a printing
chronograph is used, as is to-day the case in the
majority of observatories: the correction, then, to be
deduced from the observations of the transits thus
noted is that of the chronograph, leaving out the
personal equation of the observer, and tlje time
which elapses between the closing of the electric-
recorder circuit corresponding to an observation and


the printing of the time. But unlike an ordinary
clock with or without contact, the printing chrono-
graph does not lend itself to direct comparisons, and
it is necessary, in order to. utilise the record it gives,
to transfer this reading to that of a time-keeping
clock, preferably a synchronised clock whose dif-
ference of reading from the chronograph should be

It will be of interest to indicate a process by
which, firstly, the true correction of the printing
chronograph can be obtained, then this correction
passed on to any one of the time-keeping clocks.
To make this clear, let us suppose we are using a
Gautier printing chronograph. It will be easy, if
another type is used, to modify the processes in
question so as to render them applicable. We
will suppose, moreover, that this chronograph is
associated, as is the case at the Paris Observatory,
with an automatic recording micrometer, so as to
consider the personal equation of the observer
negligible. The working of the apparatus is then
as follows : — The movable wire of the micrometer is
kept constantly pointed to the star, and each time
it passes a fixed position an automatic contact is
produced which closes the electric-recorder circuit.

This circuit being put into action, attracts a plate
forming one arm of a lever ; the other arm is formed
of spring bands carrying at their ends points which
strike a sharp blow on a roll of paper and press it
against three wheels making a turn in one second,


one minute, and one hour respectively. Each of
these wheels bears characters in relief which print
on the parts of the roll where the points press, so
that the hundredth of a second, the second, and
the minute of the chronograph are obtained for
each transit of the micrometer wire, and therefore
of the passing of a fixed position by the star.

The wheel bearing hundredths of a second is
subjected to the synchronising action of a special
clock, which is itself synchronised by means of
relays with a time-keeping clock. Its rotary speed
is regulated to a little more than one turn per
second ; but an arrangement controlled by a second
recorder worked by the electric current traversing
the contact of the special clock, stops the wheel at
each revolution so as to reduce its speed by one
turn. It will be understood without going into
further details, that the reading determined with
the times of transit thus registered, calls for some
correction before it can be considered with the
true reading of the chronograph.

Firstly, between the moment when the transit of
a certain known position by the micrometer wire
produces an electric contact, and that when the
point presses the roll on to the ' ' hundredths "
wheel and prints a figure on it, a certain period of
time elapses. This retardation which is due to the
electrical and mechanical inertias of the electric
recorder and its plate, includes the time necessary
for the point to get into motion and the duration of


its displacement from the position of rest until it
touches the wheel.

To measure this interval of time, a process an-
alogous to that described on p. 49 is employed.

For this replace the electrical contact of the
recording micrometer by that of a clock P^ ; but
as the strength of current necessary to work the
electro-writer E is rather great and might deteriorate
the clock contacts, it is as well to interpose a relay
R (fig. 23).^ Also glue a piece of thin tin-foil to
the roll of paper a. A primary commutator, Cj
enables the contact of the relay R to be connected
at will either to the electric-writer E and its battery
/, or to a second circuit comprising a commutator
C2, a battery p^, a condenser K^ provided with a
resistance, and finally the primary of a small induc-
tion coil Bj. The condenser has a value of 2 mfds. ,
for example, and the resistance joined across its
armatures — about 25,000 ohms. The primary of
the coil Bj with its core can be moved inside the
secondary so that the induction can be varied at will.

The body of the chronograph, and in particular
the ' ' hundredths " wheel r, as well as the tin-foil a,
are all joined to the free terminals of the com-
mutator Cg.

A second clock Pg whose period is regulated so as
to differ by about ^-J-g of a second from that of Pj

' Clock and relay may consist of the synchronising clock and recorder
of the chronograph if this recorder has a contact such that it can be
used as a relay.



is provided with an electrical contact put in circuit
with a battery /g, a condenser K^ and its resistance,
and finally with the primary of a second induction

E :E




mmmm mwm



Fig. 23.



coil Bg. The secondaries of the two coils are joined
in series with a telephone T.

The two commutators being pressed down to the
right, observe by means of the telephone T the
coincidences of the beats of the relay R and those
of the clock P, ; measure their intervals and note



the time recorded by Pj at the moment of the last
coincidence observed.

At this moment move the two commutators to
the left and note the time recorded by Pg at the
moment of the first coincidence observed (in turn)
between the beats of Pj and those of the contacts of
the foil a with r. It is then easy to deduce, as
indicated in a previous example (p. 50), the interval
of time which separates the moment of a contact of







Fig. 24.

the relay R from the moment of the consecutive
contact of a with r. This interval is evidently the
same as that which exists when the electric-writer is
controlled either by the recording micrometer or
by a contact made by hand with a weight. If it is
possible to arrange in front of the plate of the
electric regulator E' of the chronograph an adjust-
able electrical contact placed in circuit (fig. 24) with
the electric-writer E and its plate, each beat will
be written on the paper roll, and it is easy to
reckon the regularity of the synchronisation of the


"hundredths" wheel since the "hundredths" figure
marked on the roll should then always be the same.
It is equally easy to verify the regularity of the
speed of the ' ' hundredths " wheel by controlling the
electric-writer, directly or through relays, by a clock
whose period differs by ^-^ of a second, for instance,
from that of the clock acting on the electric-regulator.
If the speed is constant, the "hundredths" figures
marked successively on the roll will differ from each
other by one unit.

If this is not the case, the speed is irregular :
the differences, however, indicate how much this
irregularity is, and so it is possible to calculate it.
Arrangements could equally be fitted up without
much difficulty, enabling the beat of a round second
of the electric-regulator, which immediately follows a
prick of the electric-writer, to be inscribed on the roll.
It would serve no useful purpose to describe these
arrangements here.

The true reading of the chronograph, i. e. that which
is obtained if the point strikes the ' ' hundredths "
wheel at the very moment the contacts of the re-
cording micrometer are made, being known, it
remains to determine that of the synchronising time-
keeping clock or of any timekeeper whatever.

P'or this it is sufficient to arrange again opposite
the plate of the electric-regulator a contact «(fig. 25)
whose position can be adjusted so that it closes at
the moment the synchronising action occurs, i.e.
when the plate is at the end of its motion, by



supposing that this moment corresponds exactly
to the full second of the chronograph, and arrange
as shown in figure 25.

Pj is a time-keeping clock, beating the same time
as that which controls the electric-synchroniser of
the chronograph and with which the latter is to be









J. 2)1



Fig. 25.
Pj is a clock whose period differsffrom

that of the former by y^^ of a second, for instance.
The electrical contacts of the three instruments are
put in circuit (each on its side) with a condenser
provided with a discharging resistance, a battery,
and the primary of an induction coil.

A commutator C permits of the successive com-
parison of the auxiliary clock Pj with the contact n


of the chronograph and the clock P^ by the method
of coincidences. Then from these successive com-
parisons with the auxiliary clock, that of the
electrical beats of P^ and the chronograph is

If the electrical contact of the clock P^ has to
close other contacts, either to directly synchronise
other clocks or to control relays, the arrangements
are rather more complicated, but easy to conceive.
There is no occasion to indicate them here. They

1 3 5 6 7 8

Online LibraryParis (France). Bureau des longitudesWireless time signals. Radio-telegraphic time and weather signals transmitted from the Eiffel Tower, and their reception → online text (page 3 of 8)