On the Discovery of the Period of the Crab Nebula Pulsar

R.V.E. Lovelace & G. Leonard Tyler, 2012, The Observatory, V. 132, p. 186

The discovery of radio pulsars (1) gave rise to an explosion of knowledgeof stars and their evolution, immediately attracting the interest and atten-tion of astronomers world-wide. It also sparked widespread public interestin pulsars. Subsequent discoveries over the next year resulted in discoveriesof similar sources, also ‘ticking’ with periods of about a second which couldbe explained by pulsations of a white dwarf. The discovery of the periodof the Crab nebula pulsar with regular emission at a 33 ms period was thefirst detection of a source with a significantly shorter period which couldbe readily explained only by a rotating neutron star. The discovery of thispulsar has not been described previously.

Richard Lovelace, then a Cornell graduate student in physics workingfor Prof. Edwin Salpeter, spent the period from April to December 1968 atthe Arecibo Observatory, doing research on the newly discovered pulsars.He was assigned by Prof. Salpeter to develop computer codes for searchingfor pulsars in digital records (2), and had developed a ‘fast-folding code’ (2),with which he discovered the 0.5579 s period of a new pulsar in Vulpecula(3).

Later that year, in September and October, after discussions with AreciboObservatory visitors Len Tyler (a research associate at Stanford Univer-sity) and Cornell Prof. Tommy Gold (who was convinced that pulsars weremagnetized rotating neutron stars—possibly rapidly rotating—rather thanpulsating white dwarfs (4)), Lovelace began work on a new code designedto find pulsars with much shorter periods. The new code was based on afloating point Fast Fourier Transform (FFT) code written by Tyler.

The Arecibo computer at the time was a Control Data Corporation CDC3200 which had a magnetic core memory of 32K words of 24 bit length.Lovelace developed an integer based FFT code using half-words (12 bitseach) which accommodated FFTs of N = 16, 384 = 214 signal samples. Fora sampling rate of ∆t, the FFTs were calculated from spans of data of du-ration T = N∆t. This long sample length made possible a fine frequencyresolution of δf = fN/8192, with fN = (2∆t)

−1 the Nyquist frequency.These choices were important for i) increasing the output signal-to-noise ra-tio, ii) for allowing a search over a wide range of pulsar periods—includingthe shorter periods expected of rotating neutron stars, and iii) for discrim-inating against local power line noise. (The latter was required because in1968 the frequency variations of the local electric power from one span ofdata to the next were always much larger than the δf of our FFT’s!) All

1

8192 values of the received signal power versus frequency were displayed ona line printer page as a folded raster scan, i.e., side-to-side in rows and fromtop-to-bottom with values 0, . . . , 9, and ‘X’ for values greater than 9. Thesignal root-mean-square value was used to renormalize the signal so that itsaverage Fourier power was ∼< 1. The integer code ran relatively fast on thisCDC computer, and for this reason it was named Gallop (5, 6).

The Arecibo computer was available to Lovelace between midnight and 8am on weekdays, and essentially full time on weekends, which was convenientsince Lovelace lived at the observatory. It was his good fortune on theovernight computer run 9–10 November 1968 to find in the FFT rasterpatterns of 196.5 MHz scans of the Crab Nebula (7) a highly significantpeak at the frequency f = ((4951 + 4952)/2)fN/8192 = 30.22 Hz, wherethe Nyquist frequency was fN = 50 Hz for these data comprising samplesat 100 Hz. The corresponding pulse period was P = 33.09 ms (5). Part ofthe line printer output of the discovery is shown in Fig. 1 of Ref. 6.

This was an exciting discovery, but for a day or two it was met withskepticism by the observatory astronomers because the observed repetitionfrequency from the pulsar was close to the second harmonic of the local 60Hz electrical power grid. As mentioned above, this was ruled out becausethe 60 Hz signal was known to change its frequency from one data span tothe next. Follow up measurements of the Crab pulsar established that thepulse period was 33.09 ms.

Observations at a 100 Hz sampling rate were Fourier analyzed using163.8 s spans of data. Effectively, the Fourier transforms stacked about 5000periods of the received pulsar signal, resulting in an improvement of thesignal-to-noise ratio by a factor of

√5000 ≈ 70.

Subsequently, the Gallop code was applied to drift-scan observations ofthe Crab nebula and this established that the pulsar’s location was near thecenter of the Crab Nebula (6).

A few days prior to our discovery of the period of the Crab pulsar, D. H.Staelin (MIT) and E. C. Reifenstein (NRAO) announced the discovery ofdispersed pulses from two sources in the vicinity (±2◦) of the Crab Nebula(7). In the paper describing the discovery (8) they report dispersed pulsesfrom the sources NP 0527 and NP 0532, characterized in the abstract bythe statement that “Both sources are sporadic, and, no periodicities are evi-dent.” Our pulsar search at Arecibo was independent the NRAO work, and itwas designed for discovering short period pulsars. The NRAO observations(8) at 110–115 MHz were made with a 0.05 s time constant which precludeda measurement of the period of the Crab pulsar and no periods was reported.Furthermore, at frequencies of 110–115 MHz interstellar scattering broad-

2

ens the Crab pulsar pulses to a width much larger than the pulse period.Fortuitously, at the 196.5 MHz frequency of the Arecibo observations thebroadening was less than the pulse period. The identification of the Crabpulsar found at Arecibo with NP 0532 found at NRAO was established bythe agreement of the dispersion measures observed at the two observatories(6).

The short period of the Crab pulsar made it very unlikely that the starwas a pulsating white dwarf. It became clear that the pulsar was a ro-tating magnetized neutron star as advocated by Prof. Gold (4). Indeed,several weeks after the discovery of the period, the ‘spin-down’ or expectedlengthening of the period was observed (9); the rotational energy releasedin the form of magnetized winds was later shown to be sufficient to powerthe electromagnetic missions of the nebula (10).

Yours faithfully,Richard V.E. Lovelace

Department of Astronomy,Cornell University,Ithaca, NY 14853

and G. Leonard TylerDepartment of Electrical Engineering,Stanford University,Stanford, CA 94305

References

(1) Hewish, A., Bell, S.J., Pilkington, J.D.H., Scott, P.F., and Collins,R.A. “Observations of a Rapidly Pulsating Radio Source”, 1968, Nature,217, 709-713.

(2) Lovelace, R.V.E., Sutton, J.M., and Salpeter, E.E., “Digital SearchMethods for Pulsars,” 1969, Nature, 222, 231-233.

(3) Craft, H.D., Sutton, J.M., and Lovelace, R.V.E., “A New Pulsar inVulpecula,” 1968, Nature, 219, 1237-1238.

(4) Gold, T., “ Rotating Neutron Stars as the Origin of Pulsating RadioSources,” Nature, 218, 731-732.

(5) Lovelace, R.V.E., Sutton, J.M., and Craft, H.D., “Pulsar NP 0532Near Crab Nebula,” IAU Astronomical Telegram Circular No. 2113, 18November 1968.

(6) Comella, J.M., Craft, H.D., Lovelace, R.V.E., Sutton, J.M., and G.Leonard Tyler, “Crab Nebula Pulsar NP 0532,” 1969, Nature, 221, 453-454.

(7) Scans of the Crab Nebula at 196.5 MHz were digitally recorded on

3

magnetic tape for later computer analysis earlier in the week by Lovelace’sArecibo collaborators H.D. Craft, J.M. Sutton, and J.M. Comella.

(8) Staelin, D.H., & Reifenstein, E.C., “Pulsating Radio Sources Near theCrab Nebula,” IAU Astronomical Telegram Circular No. 2110, 6 November1968.

(9) Staelin, D.H., & Reifenstein, E.C., “Pulsating Radio Sources nearthe Crab Nebula,” 1968, Science, 162, 1481-1483.

(10) Richards, D., “NP 0532”, IAU Astronomical Telegram Circular No.2114, 25 November 1968.

(11) Gold, T., “Rotating Neutron Stars and the Nature of Pulsars,”1969, Nature, 221, 25-27.

4

No. 2110

PULSATING RADIO SOURCES NEAR CRAB NEBULA •

W. E. Howard, III, National Radio Astronomy Observatory, Green Bank, West Virginia, transmits the following information from D. H. Staelin and E. C. Reifenstein, III: "Two pulsating radio sources have been found with the 300 ft. transit antenna of the National Radio Astronomy Observatory. These sources, tentatively designated NP 0527 and NP 0532 pending the determination of more accurate pos-itions, are both located in the vicinity of the Crab Nebula and could be coincident with it. Each source was observed on three days in October, 1968 with a 50 channel receiver covering the band 110-115 MHz with 0.1 MHz resolution. Both sources are so sporadic that no periodicities are evident. The pulse width appears less than 120 ins. The maximum observed pulse energies for NP 0527 and NP 0532 were 207 x 10 -26 and 16 x 10 -26 (jm-2 Hz -1 ), respectively. Other source parameters are:

NAME c=1 ,,IY) 6 1950 Nedi cm-2 Period

NP 0527 NP 0532

5h27

m±6m

5 32 ±3 +22'30'71° +22 30 1'2

1.58±0.03 x 1020

1.74±0.02 x 1020

Science, New Series, Vol. 162, No. 3861 (Dec. 27, 1968), pp. 1481-1483

Pulsating Radio Sources near the Crab Nebula

Abstract. Two new pulsating radio sources, designated NP 0527 and NP

0532, were found near the Crab Nebula and could be coincident with it. Both sources are sporadic, and no periodici-

ties are evident. The pulse dispersions indicate that 1.58 ± 0.03 and 1.74 ±

0.02 x 10" electrons per square centi-

meter lie in the direction of NP 0527

and NP 0532, respectively.

During a general search for dispersed periodic signals, an unusual pair of pulsating sources was found in the vi-cinity of the Crab Nebula. They have been tentatively designated NP 0527 and NP 0532, pending more accurate measurements of position. The search was prompted by the discovery of peri-odic pulsed radio sources by Hewish et al. (1). The antenna was the 300-foot (90-m) transit telescope of the National Radio Astronomy Observa-tory (2), Green Bank, West Virginia. The receiver monitored the band be-tween 110 and 115 Mhz with 50 chan-nels, each 0.1 Mhz wide. The time constant of the receiver was 0.05 sec-ond, and all channels were sampled every 0.6 second. The output was re-corded in digital form on magnetic tape.

The sources were observed during normal drift scans on 17, 19, and 21 October 1968. The beam width of the antenna (2.5°) permitted observations for 15 minutes each day. The sources become more evident when displayed in a time-frequency diagram (Fig. 1). A series of four pulses from NP 0532 and a single pulse from NP 0527 are shown; both the linear and second-order terms of the dispersion relation and also some narrow-band interfer-ence can be seen.

The apparent variations in pulse duration indicate variations in pulse strength and result primarily from the 50-cosec time constant and the thresh-old method of display. The absence of strong variations with frequency, de-spite the strong variations with time, is one of the unusual characteristics of these sources, and it suggests that most of the variations may originate in the

immediate vicinity of the sources. The illustrated pulses were observed with circular polarization to minimize the effects of Faraday rotation.

The stability of the pulse dispersion 27 DECEMBER 1968 1481

110o 4 1 3 6 5 2

115 Fr

equ

enc

y (

Mhz

)

114

113

112

• ." t*

A

°t3t3°

O 0

00000

0 0 0 0

s is

0 • •

0 0

'f4 00 : 00

. 8 °P.

l it 1 2

• 206 50—

Ene

rgy

(10'j

oule

hz' m

-7 )

40—

0

0 0 0

0

30 —

0

0

02

00

20—

O o

°° ° 0

0 • 0 0 0

° ° 0

••• 0

O

co 00

1 0 - 00 8 • 0

50

.1 1.50 1.60 1.70 30 I"40 Electrons/cm' x 10"

Fig. 2. Number of observed pulses at each value of dispersion. The lower dispersion peak corresponds to NP 0527; the other corresponds to NP 0532.

1482

1 '90 1.80 rni4-R

114 40

30

20

10

0

Time (sec)

Time (sec)

Fig. 1. Time-frequency diagram of radio pulses observed with circular polarization. (A) One strong and three weak pulses received from NP 0532 on 21 October 1968; (B) one typical pulse received from NP 0527 on 19 October 1968. Open squares and closed circles represent deviations from the mean of 4.2 and 8.3 standard deviations, respectively.

is indicated in the histogram (Fig. 2).

The smaller maximum appears to be

real. The separation between the two

peaks is several times the experimental

error, and pulses at both values of the

dispersion were observed on all 3 days.

The dispersion was determined for each

pulse with a root-mean-square accu-

racy corresponding to less than 0.03 x

10 2" electron cm -2 . The centers of the

two peaks correspond to 1.58 ± 0.03

and 1.74 t 0.02 x 10 20 electron cm -2 .

Two values of dispersion could orig-

inate either from one source which has

two emission modes with different ray

paths, or from two separate sources.

Under the assumption that the two

sources are separate, they have tenta-

tively been designated NP 0527 and

NP 0532; the latter source is more ac-

tive and more dispersed. The observed

1950.0 positions of NP 0527 are a

5" 27"' ±6'", 8 = 22°30' ± 2°; those

of NP 0532 are a = 5" 32'" ± 3m,

S = 22°30' ± 2°. The positions of both sources could be coincident with

the Crab Nebula, located at a = 5" 31'",

8 = 21°58'. An association of a pulsat-

ing radio source with a known celestial

object would be very informative, and

the determination of a more precise

position for these sources is very im-

portant. The small angular size of the

Crab Nebula would make accidental

coincidence of the positions of the

sources with that of the nebula highly

unlikely. Such an association would

support the view that pulsating radio

sources may be neutron stars formed

in explosions of supernovas (3). Scin-

tillation mechanisms may also be re-

sponsible. If one or both of these

sources are within the nebula, then

still more precise measurements of po-

sition may permit one to ascertain

whether there are any associations with

the x-ray source (4), small low-fre-quency radio source (5), or central

stars.

The observed dispersion of the pul-

sating sources is consistent with the

1

I 1 t I 1 I 1 1 1 I i

05 h 25" C5 h 30'" 05 h 35 m 05 ° 40'

Right ascension 1968.8

Fig. 3. Pulse energies of all pulses observed on 21 October 1968. The triangles and circles represent pulses from NP 0527 and NP 0532, respectively. The dashed line represents a detection threshold which varies with receiver noise and interference, and which was chosen to be approximately 5 standard deviations above the mean.

SCIENCE, VOL. 162

estimated distance to the Crab Nebula of 2 kparsec (6). If the sources are at this distance, then the average electron density in the interstellar medium is < 0.028 electron cm -8 (7). If most of the electrons are close to the sources, then the interstellar electron density is even less.

The time variations of the sources are also unusual. The pulses tend to be isolated, although in NP 0532 they have been observed as close together as 0.21 second; NP 0527 is further distinguished by the brevity of its ac-tive periods, generally lasting no more than a few minutes during the observ-ing period (15 minutes).

If the sources each have a unique period, then these periods must be less than 0.25 and 0.13 second for NP 0527 and NP 0532, respectively. These limits were determined by checking whether the time intervals between adjacent pulses were consistent with any single period. The upper limits on the source periods were constrained by the time resolution of the observations. If ob-servations with greater time resolution

fail to yield unique periods for these sources, they then constitute a new class of radio sources distinct from the highly periodic pulsars. In Fig. 3 are shown those pulses that were observed on 21 October 1968. No pattern is evident.

DAVID H. STAELIN * EDWARD C. REIFENSTEIN III

National Radio Astronomy

Observatory, Edgemont Road, Charlottesville, Virginia

References and Notes

1. A, Hewish, S. J. Bell, J. D. H. Pilkington, P. F. Scott, R. A. Collins, Nature 217, 709 (1968).

2. Operated by Associated Universities, Inc., under tract with NSF.

3. M. J. Large, A. E. Vaughan, B. Y. Mills, Nature 220, 340 (1968).

4. S. Bowyer, E. T. Bryam, T. A. Chubb, H. Friedman, !bid. 201, 1307 (1964).

5. A. Hewish and S. E. Okoye, INA. 207, 59 (1965).

6. V. Trimble, Astron. J. 73, 535 (1968). 7. H. I. Habing and S. R. Pottasch, Astrophys.

J. Left. 149, L119 (1967). 8. We are grateful for the assistance of W. D.

Brundage and K. A. Braly and for the support of the National Radio Astronomy Observatory.

• On leave of absence from the Department of Electrical Engineering, Massachusetts Institute of Technology.

27 November 1968 ■

27 DECEMBER 1968

1483

IAU.No.2110.pdfPage 1

IAU.No.2110.pdfPage 1

Science,V162 Dec 27 1968.pdfPage 1Page 2Page 3