___|^^^_
ii''il'!' i: '" ■ il'li 1 ' Wi'ufyj I'A'M tllii ii'!"iljl nil l' ! i' ,; lii i 1. lll'li Wn'i li.
VjkH
WwMmMwwvmTawfmwMn. 9MB tm
i''i ! i!i ; ' ifwiii Mii'i i' ,; iii'il i Si* ii'iiiiwli , i' ! '! WsM v'i iiii'ls' !W;™fl 'i'
i
1 ..
i'li'i'iifri'fi-lii 1 :'!'!. |i. , !i' : ■!
t ; ttlll!i ; i''!'"Hm^''- , : :i !: : ■'■.•■' ! '';^ :: ' ; : : ''|ti!irfl ., : 'i -';!^ 'l^is'^v.ijM:^! ^'''^ ■ :^ : i''-' :')!!;:.''■■ T ■ /■■ •' -■■ ; . ■ ' : ^ ' : ■■ • ' Jill ffiiiP'jil
fxMi]' |i]|™'] : ': SjjP p ■ ',; :' ' : iffllllf fflliPffl ffi ' "i'l : ' Il'li'!! nlill! K : . ' ' P > ' •, 3
'■' i • M ll 'ill ll liilil ll' '1! :K : ':i'i! I Li ill Ullil wH 1 Hi
||: i : ?? 1
,!;j',i i||i|||i!l;i ili'jftlii'l'lj
l'i|| il'li |ii)'! 'I'l'ilil'
i
kJ
illi ; 'li i'i!i!i|'Pi 1 I'i |ii'|!''lii
■■■I^HIiiiiil'i liililillif II '"■ ' ' ■'"■ Nil III 1 ill ■ 1 I'l^llll'' 1 '! I
; llii iHiil ■; .lilii - '■ ■' I! 'uiiLltlii
; "
iBIll 111; aMITO
fell mm
J
•^ ^ 2,m~*A~»- <**
Carnegie
Institution
OF WASHINGTON
Year Book 75
1975-1976
Library of Congress Catalog Card Number 3-16716
Champion Press, Inc., Baltimore, Maryland
Issued December 1976
Contents
Officers and Staff
Report of the President
Reports of Departments and Special Studies
Department of Embryology
Department of Terrestrial Magnetism
Hale Observatories
Department of Plant Biology
Geophysical Laboratory
Developmental Biology Research Group
Bibliography
Administrative Reports
Report of the Executive Committee
Abstract of Minutes of the Seventy-Eighth Meeting of the
Board of Trustees
Financial Statement
Ten-Year Financial Summary
Report of Independent Public Accountants
Articles of Incorporation
By-Laws of the Institution
Index of Names
v
1
1
3
105
273
347
485
859
961
963
965
967
971
973
974
989
993
999
President and Trustees
PRESIDENT
Philip H. Abelson
BOARD OF TRUSTEES
William McChesney Martin, Jr.
Chairman
Frank Stanton
Vice-Chairman
William T. Golden
Secretary
Robert 0. Anderson 1
J. Paul Austin 1
Lewis M. Branscomb
John T. Connor
John Diebold
Michael Ference, Jr.
Carl J. Gilbert
Hanna H. Gray
Crawford H. Greenewalt
William C. Greenough
Caryl P. Haskins
William R. Hewlett
Henry S. Morgan
William I. Myers 2
Walter H. Page
Robert M. Pennoyer
Richard S. Perkins
William M. Roth
Robert C. Seamans, Jr.
Charles H. Townes
Juan T. Trippe
James N. White
Garrison Norton
Charles P. Taft
Trustees Emeriti
1 Elected April 1976.
2 Died January 30, 1976.
Former Presidents and Trustees
PRESIDENTS
1902-1904
Daniel Coit Gilman
Robert Simpson Woodward 1904-1920
Caryl P. Haskins
T
Alexander Agassiz 1904-05
Lord Ashby of Brandon 1967—74
George J. Baldwin 1925-27
Thomas Barbour 1934-46
James F. Bell 1935-61
John S. Billings 1902-13
Robert Woods Bliss 1936-62
Amory H. Bradford 1959-72
Lindsay Bradford 1940-58
Omar N. Bradley 1948-69
Robert S. Brookings 1910-29
Vannevar Bush 1958-71
John L. Cadwalader 1903-14
William W. Campbell 1929-38
John J. Carty 1916-32
Whitefoord R. Cole 1925-34
Frederic A. Delano 1927-49
Cleveland H. Dodge 1903-23
William E. Dodge 1902-03
Charles P. Fenner 1914-24
Homer L. Ferguson 1927-52
Simon Flexner 1910-14
W. Cameron Forbes 1920-55
James Forrestal 1948-49
William N. Frew 1902-15
Lyman J. Gage 1902-12
Walter S. Gifford 1931-66
Cass Gilbert 1924-34
Frederick H. Gillett 1924-35
Daniel C. Gilman 1902-08
Patrick E. Haggerty 1974-75
John Hay 1902-05
Barklie McKee Henry 1949-66
Myron T. Herrick 1915-29
Abram S. Hewitt 1902-03
Henry L. Higginson 1902-19
Ethan A. Hitchcock 1902-09
Henry Hitchcock 1902
Herbert Hoover 1920-49
William Wirt Howe 1903-09
Charles L. Hutchinson 1902-04
Walter A. Jessup 1938-44
Frank B. Jewett 1933-49
Samuel P. Langley 1904-06
Ernest O. Lawrence 1944-58
Charles A. Lindbergh 1934-39
William Lindsay 1902-09
Henry Cabot Lodge 1914-24
Alfred L. Loomis 1934-73
Under the original charter, from the date of organization until April 28, 1904, the following
were ex officio members of the Board of Trustees : the President of the United States, the Presi-
dent of the Senate, the Speaker of the House of Representatives, the Secretary of the
Smithsonian Institution, and the President of the National Academv of Sciences.
John Campbell Merriam
1921-1938
Vannevar Bush
1939-1955
1956-1971
Robert A. Lovett
1948-71
Seth Low
1902-16
Wavne MacVeagh
1902-07
Keith S. McHugh
1950-74
Andrew W. Mellon
1924-37
John Campbell Merriam
1921-38
Margaret Carnegie Miller
1955-67
Roswell Miller
1933-55
Darius 0. Mills
1902-09
S. Weir Mitchell
1902-14
Andrew J. Montague
1907-35
William W. Morrow
1902-29
Seeley G. Mudd
1940-68
William I. Myers
1948-76
William Church Osborn
1927-34
James Parmelee
1917-31
Wm. Barclay Parsons
1907-32
Stewart Paton
1916-42
George W. Pepper
1914-19
John J. Pershing
1930-43
Henning W. Prentis, Jr.
1942-59
Henry S. Pritchett
1906-36
Gordon S. Rentschler
1946-48
David Rockefeller
1952-56
Elihu Root
1902-37
Elihu Root, Jr.
1937-67
Julius Rosenwald
1929-31
William W. Rubey
1962-74
Martin A. Ryerson
1908-28
Henry R. Shepley
1937-62
Theobald Smith
1914-34
John C. Spooner
1902-07
William Benson Storey
1924-39
Richard P. Strong
1934-48
William H. Taft
1906-15
William S. Thayer
1929-32
James W. Wadsworth
1932-52
Charles D. Walcott
1902-27
Frederic C. Walcott
1931-48
Henry P. Walcott
1910-24
Lewis H. Weed
1935-52
William H. Welch
1906-34
Andrew D. White
1902-16
Edward D. White
1902-03
Henry White
1913-27
George W. Wickersham
1909-36
Robert E. Wilson
1953-64
Robert S. Woodward
1905-24
Carroll D. Wright
1902-08
OFFICE OF ADMINISTRATION
1530 P Street, N .W '., Washington, D.C. 20005
Philip H. Abelson President
James W. Boise Bursar ; Secretary-Treasurer, Retirement Trust;
Executive Secretary to the Finance Committee
Marjorie H. Walburn Assistant to the President
Sheila A. McGough Publications Officer; Editor
Kenneth R. Henard Assistant Bursar; Assistant Treasurer,
Retirement Trust
Joseph M. S. Haraburda Assistant to the Bursar
Richard E. Hewitt Administrative Officer for Services
Marshall Hornblower Counsel
STAFF MEMBERS IN SPECIAL SUBJECT AREAS
Ellis T. Bolton
Roy J. Britten
Tatiana Proskouriakoff
DISTINGUISHED SERVICE MEMBER IN SPECIAL SUBJECT AREA
Barbara McClintock
RESEARCH ASSOCIATE AT LARGE
Harry E. D. Pollock
Carnegie Institution of Washington adheres in all phases of its operations,
including employment and educational programs, to a policy barring discrimina-
tion on the basis of race, religion, color, national or ethnic origin, or sex. In its
educational programs it admits qualified students as fellows without regard to
race, religion, color, national or ethnic origin, or sex to « all the rights, privileges,
programs, and activities generally accorded or made available to fellows at the
Institution. It does not discriminate on the basis of race, religion, color, national
or ethnic origin, or sex in administration of its educational policies, admissions
policies, fellowship programs, and other Institution-administered programs.
Report of
the President
ELECTRONICS REVOLUTION
Earlier in the century, this country experienced a long era of sustained growth
in many aspects. There was a steady increase in level of education, life expectancy,
and standard of living. Growth of all kinds was welcomed, including industrial
expansion and population increase. A feeling of progress, of achievement, of well-
being was everywhere. As a corollary, a striving for excellence and the search
for understanding were widely admired.
Today the mood of this country has turned pessimistic and negative. Those
who are so inclined can find much evidence to support these views. Growth in
the use of energy in the form of oil was suddenly curtailed in 1974. Growth in
consumption of the kind seen in the 1950s and 1960s will not occur again. That
part of the standard of living that is based on large-scale consumption of energy
is not likely to improve during this century.
However, those who prefer optimism have reason for hope. Human ingenuity
in solving problems is great. And native intelligence has been amplified enor-
mously by the use of knowledge accumulated through research. An important
product of research and a basis for hoping for a bright new future is the vitality
of the continuing electronics revolution. This revolution has been in progress for
about 60 years. Lately its tempo has increased greatly. Until recently its im-
portance was overshadowed by changes brought about by the large-scale expan-
sion in the use of energy. But the electronics revolution promises to be more
important, of more enduring consequence, than the earlier Industrial Revolution.
Some of the great changes brought about by this revolution have gone on
comparatively unnoticed ; at the start they were evolutionary rather than sudden
and drastic. The telephone, which we take for granted, was invented 100 years
ago. Nearly every decade since then, the quality and scope of service have steadily
improved while the cost (measured in constant dollars) is now a tiny fraction
of what it was 50 years ago.
Numerous applications of electronics gradually affected individuals and almost
every component and activity of society. Americans listen to commercial radio
an average of nearly four hours daily. Television has come to most homes.
During the last few years the accelerated tempo of change has led to an increased
impact on society. For example, there is the rapidly burgeoning acceptance of
the citizen's band radio, the worldwide use of the telephone, and the current
astonishingly low prices at which hand-held computers and electronic watches
are being sold. Less evident to the individual but in total more important to
society are new and increasing applications of electronics to process control for
industry, to research equipment, to merchandising, to banking, to management, to
national defense, and to a host of other activities.
This revolution, which is destined to have great long-term consequences, is
quite different in nature from the earlier Industrial Revolution. The Industrial
Revolution was based on the profligate use of energy (mainly fossil fuels) . Much
of its technology was crude, with only a modest scientific or theoretical base. In
large measure what the Industrial Revolution did was to make available and to
employ large amounts of mechanical energy. History may well describe the past
century as the Era of Brute Force.
In contrast, the electronics revolution represents one of the most noble intel-
lectual achievements of mankind. Its development has been the product of the
most advanced science, technology, and management. In many applications
electronics requires little energy. Indeed, one factor that guarantees enduring
impact for the electronic revolution is that it is sparing of energy and materials.
3
4 CARNEGIE INSTITUTION
With electronics one can control the disposition of large amounts of energy
and force, but the relationship is much like the use of the brain in directing the
effort of muscles. In some aspects, electronics can be more subtle, more nimble,
more dependable than the brain. In other applications, electronics serves as a
great extender of human capabilities by rapidly carrying out routine but complex
calculations, thus freeing the mind to make intuitive judgments and short cuts
to new insights.
In laboratories devoted to extending the electronics revolution, the use of
powerful investigative tools based on electronics is speeding new developments.
Ultimately, there must be limits to what can be done. However, the barriers are
not yet obvious, and it is clear that the revolution has some distance to go. What
has been accomplished and something of what is possible may be found in the
history and potential of certain solid-state devices. Writing about integrated
circuits formed on silicon chips, Robert N. Noyce has commented:
The cost reductions resulting from higher levels of integration have
been remarkable. The individual diffused transistor sold for a price of
approximately one dollar in 1961, a few years after its introduction.
Today, a 1024 bit memory including over six thousand transistors sells
for about one dollar a few years after its introduction, representing a
cost reduction of about 6000:1 in 15 years. The resulting effect on
equipment costs can be seen in the electronic calculator, which has been
reduced in cost by a factor of 500:1 in the last 8 years, creating a
substantial new market unforeseen only 10 years ago.
Noyce and others are confident that the performance of silicon chips will be
improved by at least a factor of 10. Already a tiny device has the computing
power that earlier required a roomful of tubes and other electronic components.
One of the factors assuring a bright future for electronics is its high degree of
social acceptance. There have been sporadic attacks on various electronic devices
such as computers, but the intensity of criticism has diminished. In comparison
to the level of criticism of chemical products or the environmental concerns
associated with nuclear and fossil fuel energy, objections to electronics have been
few. Broadly available items such as the hand-held computer, electronic watch,
and citizen's band radio enhance the public's feeling of participating in the
benefits of electronics while not bringing with them discernible side effects.
In future, electronics will provide many new tools for humanity.
There are, of course, limitations to what electronics can do. It cannot supply
the basic human needs of food, clothing, and shelter. But it can increase the
efficiency of the processes by which these requirements are met. It can contribute
to economies in the use of energy, for example, by improving the engine efficiency
of automobiles through ignition control. Better electronic communications will
lessen the need for business travel. Solar panels of improved cost and efficiency
may be developed, making them an important energy source.
Just as electronics is increasingly changing the ways in which industry and
commerce are conducted, it has also altered and continues to change the tools
for scientific research, sometimes drastically, and the new tools are creating great
research opportunities.
The use of electronic instrumentation has very much affected the conduct of
research at the Carnegie Institution. In astronomy, it has increased the volume
of the measurable universe by as much as a factor of 50. In the physical and
biological sciences, electronics has expanded the time horizon and vastly increased
the sensitivity and accuracy of many measurements. Human response time is of
REPORT OF THE PRESIDENT . O
the order of 1/10 to 1/5 second, but electronics enables humans to conduct
experiments with time spans as short as a picosecond (10 -12 second).
Two examples may be cited to illustrate the increased sensitivity and ease
of analytical determinations. At one time the analysis of a silicate rock required
a sample of about a gram and the processing occupied about a week's time. A
skilled chemist might complete a dozen determinations in a week. Today, using
the electron microprobe, one can analyze a sample of perhaps 10 -10 gram in about
10 minutes.
Another example involves the analysis of a mixture of organic chemicals such
as those found in petroleum. The original work to determine the composition
of a crude oil, the Ponca City crude, required the fractionation of many barrels
of oil — about a million grams — and efforts of many workers extending over about
two decades. Today such an analysis can be performed in a few hours em-
ploying a sample containing a tenth of a gram.
Of comparable importance is the use of small but powerful on-line computers.
These greatly facilitate the accumulation and interpretation of data and free
investigators from dull bookkeeping tasks.
Many of these tools are production models of major instrument makers such
as Hewlett-Packard. But leaders in research do not necessarily wait for some one
else to produce a new device. Indeed, they often become leaders by being first
to develop and exploit a new tool. Developing new tools and using them has
been a tradition and practice at Carnegie installations.
In the past, most new research instruments used at the departments of the
Institution were totally designed by them. The current practice is to combine
commercially available electronic packages such as amplifiers or minicomputers
with home-designed components to make a new device with unique capabilities.
Any account of the electronics revolution, however brief, would be incomplete
without mention of the key role that three Carnegie trustees have had and are
continuing to have in it. William Hewlett's company has made enormous con-
tributions. Scientists and engineers worldwide are impressed with the quality
and usefulness of the many instruments that Hewlett-Packard has innovated.
Lewis Branscomb has had a long association with electronics, and as Vice-
President and Chief Scientist for IBM he is helping guide the development of
new products for that company. Charles Townes, who received the Nobel Prize
for his research in quantum electronics that led to the maser-laser principle,
continues creative work. His current radio telescope studies of molecules in space
and his infrared work are advancing the frontiers of astronomy.
EVOLUTION OF THE INSTITUTION
On the surface the Institution appears much as it did six years ago. The five
major departments retain their names and most of the staff members. However,
there have been substantial changes. Four new directors have taken over, research
programs have evolved, new equipment has been obtained, and major facilities
have been constructed.
Department of Plant Biology
The changes at Plant Biology are particularly satisfying. Under Winslow R.
Briggs the department has become what some consider the world's best center
for plant research. He has built on solid foundations laid by his predecessor.
For many years, C. Stacy French had conducted a small though well-regarded
department. During his tenure, French attracted fellows and guest investigators,
6 CARNEGIE INSTITUTION
many of whom were or became leaders in the field. He also provided leadership
in creation and use of new instruments for plant research.
As the time for French's retirement approached, special thought was given to
the opportunities inherent in plant biology. It was obvious that advances in plant
biology might provide great benefits to a world seeking solutions to its problems
of food and energy. The great agricultural research stations could be expected
to conduct excellent studies on existing crops. For Carnegie, the optimal oppor-
tunities for research lay in long-term fundamental studies of plant biology that
carry with them the potential for breakthroughs. Such new knowledge could be
expected ultimately to lead the way to better plants and to better management
of existing species.
For some years the biology of plants did not enjoy so intense a study as that
devoted to other related sciences. The substantial funds available for medically
oriented research had led to great advances in molecular biology and biochemistry.
The emphasis was on studies of microorganisms or animals. But these large-scale
efforts developed concepts, techniques, and powerful equipment applicable to
plant research.
Briggs was fully aware of the possibilities. He was attracted also to the
abundant opportunities for interaction with biology professors and students at
Stanford University. It was clear, though, that to create an establishment com-
mensurate with the opportunities would require extensive renovation of the
laboratories and new construction. The necessary planning, design, and construc-
tion were completed with minimal disruption of. research efforts. The resultant
modern facilities are functional, versatile, and attractive to staff, fellows, and
guest investigators. In addition to standard laboratory equipment for biochemical
studies, the department has controlled plant growth chambers capable of main-
taining a broad range of environments. The department shares a mobile labora-
tory with Stanford (complete with minicomputer) that facilitates in situ studies
in diverse climatic areas. Special equipment is available at the laboratory for
observations on the photosynthetic processes at time spans as short as 10 -6
second. Other equipment is designed to measure the absorption spectra or action
spectra of intact plant materials or of membrane preparations. The output of
such devices is electronic, and they arc programmed and their output is processed
by an in-house computer that has time-sharing capabilities.
The research program at the department includes major activities that con-
tinue from the past — studies of the mechanisms of photosynthesis and physio-
logical ecology. Briggs has brought a new program involving himself and others
in the study of membranes. Membranes have a role in photosynthesis, and the
temperature behavior of lipids of membranes is relevant to physiological ecology.
Thus the three major thrusts of the department interact well together.
One measure of the overall quality of the department is its attractiveness to
Stanford biologists and to guest investigators from elsewhere. In his current
annual report Briggs lists the many people who are now finding the department
a satisfying place for scholarship.
Hale Observatories
Astronomers of the Hale Observatories continue to be pathfinders in the
study of the universe and its contents. They are concerned with providing
answers to basic questions: the age of the universe, the formation and evolution
of galaxies and stars, and the nature of bizarre physical processes in x-ray
sources, neutron stars, and black holes. These profound problems have justified
the Institution's largest construction project in several decades — the new observa-
REPORT OF THE PRESIDENT /
tory at Las Campanas with its 2.5-meter Irenee du Pont telescope. Staff astron-
omers are at the forefront in developing new instruments for measurement of
the faintest and most remote objects that can be detected. Because optical
astronomy can provide enormous amounts of detailed information about chemical
composition, temperature, pressure, density, strength of gravity, electric fields,
and magnetic fields in particular regions of stars, earth-based telescopes will
continue to have a central role. Hale staff have interacted with observers studying
other portions of the electromagnetic spectrum. They have also been leaders in
developments that have extended the reach of telescopes.
New electronic devices detect and count individual photons. The ability of
telescopes to detect dim and very distant objects is limited by a faint but ever-
present airglow in the earth's atmosphere: That is, even when the telescope is
pointed to areas of the sky where there are no stars, some light is received. With
electronic detectors, it is possible to subtract the airglow effect when spectra
are being recorded. The use of such electronic detectors extends the distance to
which humans can measure the universe. By such means it will be possible to
detect light from an object so distant that the light that reaches us began its
journey 15 billion years ago.
For the Hale Observatories, one of the big developments of the past six years
has been the creation of a major observatory in the southern hemisphere.
The Mount Wilson Observatory of the Carnegie Institution was the world's
leader in astronomy for nearly four decades with its 1.5-meter and 2.5-meter
telescopes. With the inauguration in 1948 of the 5-meter telescope on Palomar
Mountain by the California Institute of Technology, a new powerful tool was
made available to Carnegie staff through an arrangement in which Mount Wilson
and Palomar were operated as a unit in what came to be called the Hale Observa-
tories. With the passage of time, the telescopes at Mount Wilson lost some of
their value because of an increase in scattered light from the metropolitan region.
If the Institution were to provide its share of facilities to the Hale Observa-
tories, it would be necessary to add something new. The southern skies held
relatively unexploited opportunities, and surveys showed that viewing conditions
in north-central Chile were particularly favorable. Initial efforts to obtain outside
support for a 5-meter telescope proved disappointing. However, a gift by
Henrietta Swope and a later gift from the Crawford Greenewalts provided
impetus to the building of an observatory on Las Campanas in the Norte Chico
district- of Chile. Facilities there include a 0.55-meter telescope owned and
operated by the David Dunlap Observatory of the University of Toronto, the
1-meter Swope telescope, and the 2.5-meter Irenee du Pont telescope.
Into the du Pont telescope have gone the products of the genius of two of
the world's most experienced designers and engineers — the optical skill of the
late Ira S. Bowen and the mechanical expertise of Bruce Rule. The 2.5-meter
du Pont telescope is smaller than, for example, the 5-meter Hale at Palomar, so
it takes longer to collect the same amount of light. But its capabilities are unique
among telescopes of its size. The du Pont telescope optics employ principles of
design that make it possible to expose photographic plates 51 centimeters square,
or 12 times the usable area of those employed with the 5-meter at Palomar. The
focal length is the same as the Hale 5-meter, so the du Pont can reach objects
as faint and distant as any other instrument on earth, although requiring more
time. But because of the enormous size of the photographic plates in the du Pont
telescope, it will be uniquely able to perform crucial astronomical surveys. It can
perform surveys three times faster than the huge 5-meter Hale telescope by scan-
ning much greater areas of the cosmos.
8 CARNEGIE INSTITUTION
Spectrograph^ observations are crucial to studies of stellar chemical composi-
tion and evolution. To provide maximum efficiency for spectrographic analysis of
light from the south polar region of the sky, a radically new coude system has
been designed that requires only three reflections of light instead of the usual five.
Since at each reflection a considerable amount of light is lost, the new design will
substantially increase the effectiveness of the instrument.
Another feature of the du Pont telescope is its electronic control system that
employs programmable microprocessors. These devices represent some of the
latest advances in electronics. The microprocessors of the telescope's electronic
control system are used to set the telescope accurately and quickly, move the
dome and the windscreen automatically, provide correct rates for driving the
telescope, and assist the astronomer with data acquisition. Automated controls
and sensing devices will permit much of the observing to be done remotely, with
the astronomer in the control room. An integrated closed-circuit television system
permits guiding on very faint objects.
With its exceptionally wide field for direct photography, the du Pont telescope
is well adapted for searching for distant quasars. It is also well adapted for
studying objects that are closer. Observers in the southern hemisphere have the
advantages of studying regions of the sky that have not been so thoroughly
examined as those of the northern. In particular, the Greater and Lesser Magel-
lanic Clouds present special opportunities. They are our nearest galactic neigh-
bors, and they contain very old stars, but they also contain very young ones.
The Magellanic Clouds are very active regions of star formation.
The new telescope is becoming operational at a time of great challenges and
opportunities for astronomy. The subjects are truly cosmic — events at the far
reaches of the universe, events at the beginning of time, the creation and evolution
of galaxies, the birth and death of stars.
Geophysical Laboratory
During the early years of this century, the easily available oil and mineral
resources were heavily exploited. Most of the oil that was formed during hundreds
of millions of years was consumed. At the beginning of this century, the copper
ore being mined contained about 3% of the metal. Currently the average grade
is about 0.7%. The degree of exhaustion of the high grades of various elements
differs, but the overall picture is similar to that of copper.
Man will continue to turn to the earth as an essential source of raw materials.
However, it will be necessary to employ much more advanced knowledge and
technology to the processes.
The Geophysical Laboratory has been a leading source of new information
about the earth. The late W. W. Rubey, one of the founders of modern geology
and a trustee of the Institution, evaluated the work of the Laboratory in an
article published in 1974 as part of a survey of major developments in earth
sciences occurring in the last 50 years. In discussing advances in knowledge about
the petrology of igneous and metamorphic rocks, Rubey cited the Geophysical
Laboratory as having made perhaps the most far-reaching contributions through
applications of principles of physical chemistry. He summarized his discussion
of the Institution's work by saying
It is probably fair to state that the science of petrology has been made
over by the results of experimental work at the Geophysical Laboratory
and by [N. L.] Bowen's theory. Perhaps an even greater contribution
of the Laboratory is one that extends beyond petrology into many
REPORT OF THE PRESIDENT i>
other fields of science. This is the demonstration that problems of
seemingly hopeless complexity may be divided into simpler, more
tractable parts and brought into the laboratory for analysis and solution.
In general terms the objectives of the Geophysical Laboratory are much the
same as they were nearly 70 years ago. However, the specific research programs
are entirely different, and they are carried out using different equipment. Indeed,
a history of the laboratory would necessarily emphasize a continuing change of
equipment, techniques, and interpretations,, with many of the new tools invented
and developed there. Especially during the past decade, there has been heavy
and increasing use of electronics to better understand the processes involved in
the formation and evolution of the earth.
Various kinds of analytical tools, such as the electron microprobe, Mossbauer
equipment, chromatography coupled with electronic detectors, spectrophotometry,
and micro x-ray crystallography, have shortened the time for determinations by
factors of a hundred or more while diminishing the size of sample required by
many orders of magnitude — in some instances by factors of more than a million.
Much of the analytical equipment utilizes dedicated minicomputers to program
the analytical procedure and to record and analyze the results. Another important
use of electronics is in the storage and retrieval and processing of the vast
amounts of information produced by numerous workers. The rocks of the earth
are many and they have been generated by a variety of processes from differing
starting materials. Over the years many tens of thousands of chemical analyses
have been performed on rocks in laboratories around the world. Information of
this sort is now so voluminous and so widely dispersed as to make its retrieval
and reduction by conventional methods grossly inefficient. Chayes has designed a
computerized information system that efficiently scans a model data base con-
taining 16,000 multicomponent analyses, and he is planning a much larger and
more sophisticated world data base for igneous petrology.
Over the years, the Geophysical Laboratory has been a leader in the invention
and application of high-pressure techniques for the study of minerals. Using a
device that permitted accurate control of pressure and temperature, Yoder made
many pioneering studies at pressures of 10 to 15 kilobars during the 1950s and
1960s. This pressure corresponds to a depth of 30 to 45 kilometers and to the
region of many rock-forming processes. In 1960 Boyd and England developed
a solid media press routinely capable of pressures of 50 kilobars. This device has
been widely copied and used. During the last several years Mao and Bell have
redesigned and extended the range of the diamond press that was originally
developed by Van Valkenberg and others at the National Bureau of Standards.
This year they were able to achieve a pressure of 1200 kilobars and to make
routine measurements at 900 kilobars. They have accordingly been able to study
behavior of minerals at pressures of the core-mantle boundary (2000-2500 km
deep).
Department op Terrestrial Magnetism
Over the years a central thread of the work of the department has been the
physics of the earth and its environs. In the early days, a major effort was de-
voted to study of the earth's magnetic field first by expeditions, including those
of nonmagnetic ships, and then at magnetic observatories. The program was of
considerable practical importance, but after the accumulation of several decades
of data, scientific interest in the work diminished. These magnetic studies were
halted when Merle Tuve returned to the department as director following his
10 CARNEGIE INSTITUTION
very successful war work with the proximity fuze. Geophysical investigations of
seismicity, radioactive dating, and cosmic rays were initiated, and some ionospheric
studies were continued. A long-enduring program in low energy nuclear physics
was carried out. In addition, a dynamic effort in biophysics was successfully
launched and maintained. Interest shifted from the ionosphere to the new science
of radio astronomy and to development of the Carnegie image tube for astro-
nomical observations.
These broad programs of the department continued to progress following
Tiive's retirement in 1966. The biophysics group made many fundamental con-
tributions to molecular biology [see Roberts in Year Book 741- Eventually,
however, as the study of molecular biology matured, the DTM's biophysics
program lost some of its distinctiveness. With reluctance the biophysics program
was phased out.
In 1975 George W. Wetherill succeeded Ellis T. Bolton, who had followed Tuve
as director. Wetherill was trained in physics, and his research had included
geophysics. His current work on physics of the origin and evolution of the solar
system fits well with other research activities in the department.
As the success of DTM's program has shown, a training in physics is particu-
larly useful as a preparation for problem solving. Since the department is for the
most part staffed by physicists, it is not surprising that they have invented,
developed, or adapted many items of electronic equipment.
For example, two developments conceived by the radio astronomy group are
still in routine use at radio observatories. The development of the multichannel
line spectrometer only two years after the discovery of emission from interstellar
neutral hydrogen enabled radio astronomers to study the entire frequency spec-
trum of a region without retuning the receiver for each frequency. The develop-
ment of a traveling feed for large transit radio telescopes greatly increased the
sensitivity of these instruments. Also, image intensifiers developed under the
Carnegie image tube program were distributed to 34 observatories, and their
worldwide use opened a new era in observational astronomy.
The DTM geophysics group has long been recognized for its development of
instrumentation. In the last ten years two significant new instruments have
allowed acquisition of much valuable new seismic data.
In the mid-1960s, Sacks designed a broad frequency band, high dynamic range
seismograph system that for the first time allowed near-source recording of long-
period seismic waves. These seismometers are now produced commercially, and
special features of the magnetic tape recorder which enable stable low-speed
recording have been copied by other seismologists as well as having been in-
corporated into commercial recorders.
Toward the end of the 1960s, Sacks, working with Dale Evertson from the
University of Texas, developed a bore hole strainmeter, and it has now been
demonstrated that this device reliably records earth strain at a sensitivity several
orders of magnitude better than other strainmeters. Major use of this instrument
is being made in the Japanese earthquake prediction program.
In the field of isotope geochemistry, the DTM group was one of the first to
make use of minicomputers for automatic control and data acquisition in the
operation of mass spectrometers. This development increased the efficiency and
reproducibility of mass spectrometric operation enormously, and permitted the
reduction of quantities of data which would otherwise have been prohibitively
large. More than anything else, this made it possible to achieve the precise isotope
ratio measurements needed to carry out the investigations of the sources of
oceanic basalts by Hart and his collaborators.
report of the president 11
Department of Embryology
Few departments have changed as completely during the tenure of a director
as did Embryology under James D. Ebert. When he came to the Institution in
1956, the department was located in crowded quarters in the Johns Hopkins
Medical School. From his predecessors Ebert inherited a world famous collection
of human embryos, a monkey colony, and a program largely geared to primate
embryology. Though the department enjoyed a high standing in its field, the
challenges inherent in the study of morphology of embryos had diminished.
Ebert gradually changed the thrust of the research and moved the department
to a new building on the Homewood Campus of Johns Hopkins. Today there
are practically no vestiges of the earlier program. While the study of development
remains a general goal, the approach to it is largely in terms of molecular biology.
A typical major effort is one devoted to study of the structure, arrangement, and
evolution of genes.
In his last report as director of the department, Ebert has described some of
the changes that occurred during his tenure. A new director, Donald D. Brown,
took over on October 1, 1976. It is too early to guess the direction in which the
department will move. But evolve it will, for the new director has drive and
enthusiasm, and his tastes are somewhat different from those of his predecessor.
Moreover, the exterior environment is changing, and with it new opportunities
will emerge.
The Year in Review
GEOPHYSICAL LABORATORY
Considerable information is available from geophysical, geochemical, peno-
logical, and geological observations at or near the surface of the earth. But one
of the most challenging and tantalizing mysteries is the nature of the deep
interior of the earth.
The Geophysical Laboratory has been particularly active in furnishing the
foundations for intelligent speculation by determining the behavior of minerals
at temperatures and pressures prevalent in the interior of the earth. During recent
years the Laboratory has been a leader in two such major programs. Boyd has
led what now amounts to a sizable international effort to exploit the potential
of nodule suites propelled upward in kimberlite pipes. These mineral assemblages,
which may contain diamonds, originated at depths of about 200 km. In this
year's report are presented the results of examination of nodules from South
Africa, Siberia, and Colorado. These and earlier such studies were facilitated by
the pioneering research at the Laboratory that produced geothermometers and
geobarometers applicable to conditions at the site of origin of diamonds.
Another major program is being conducted by Mao and Bell in their experi-
ments with minerals under great pressures (about 1000 kilobars) in their new
diamond cell apparatus. One of the first steps in developing a physical model of
the earth's deep mantle and core is to determine the physical properties of
minerals and mineral assemblages likely to be there. Among the principal candi-
dates are close-packed oxides and metals including Si0 2 , MgO, iron oxides, and
metallic iron. For work employing the new cell, MgO and eFe were chosen.
Both phases had been studied earlier by shock wave techniques, and it was useful
to correlate those results with findings from static pressure measurements. Com-
pressibility of MgO and eFe was measured in the new study at 900 kilobars. It
was found that no phase changes occurred above 130 kilobars. There was agree-
ment between the shock wave and static results. The data serve to set constraints
on potential mineral properties to an equivalent depth in the earth of 2000 to
2500 km.
Exploration for ore minerals in 1900 was a simple matter. Prospectors roamed
the hills and inspected outcrops of rocks. Later, economic geologists established
empirical procedures based on observations that better guided deeper exploration.
13
14 CARNEGIE INSTITUTION
They observed, for example, that many ores are found in relation to geologic
structures. They also noted that such ores seemed to have been emplaced as a
result of motion of fluids through permeable regions associated with breakages
in the rock. A record of processes that have occurred is left in the rocks, because
the fluids react chemically during passage. The nature of the reaction is, of course,
governed by the chemical content of the fluids and the physical conditions at the
time of the reaction. Economic geologists have long used the products of such
reactions as a guide for seeking ore, but there has been no solid theoretical base
for systematizing field observations.
This year Frantz and Mao made considerable progress toward such a theoreti-
cal framework. They developed a general mathematical model to predict quanti-
tatively the mineral sequences that occur by mechanisms known as intergranular
diffusion and infiltration, and this year they solved the equations describing the
thickness of zones and the relative directions of movement of the zone boundaries.
Of particular importance, their theory aids in deducing the direction of transport
of the various species in solution. The model accommodates variable permeabili-
ties, diffusion constants, infiltration rates, and any number of components.
The model has yet to be applied to natural ore deposits; but they did show
its application to the system CaO-MgO-Si0 2 -H 2 0-C0 2 . The mineral zone se-
quences were found to be entirely dependent on the mobilities of the various
diffusing species. The lack of data on diffusion constants will, for the moment,
retard application of the theory. Eventually, however, it will be necessary to
observe only the minerals in the end zones and the thickness of the zones. By
combining these data with the diffusion constants determined in the laboratory,
it will be possible to predict the mineral content and thickness of the intermediate
zones.
The behavior of minerals containing the abundant oxides (Si0 2 , A1 2 3 , MgO,
Na 2 0, CaO, and FeO) is extremely important to the understanding of earth
processes. These common minerals have valuable but limited usefulness as indi-
cators of earlier events in the evolution of igneous rocks. A basaltic lava can
arise from a broad range of original ingredients. In contrast, trace elements can,
in principle, provide much sharper diagnostic tools and they are also of interest
because of potential economic value. The behavior of a given trace element is
partly governed by the major components of the rocks or magmas with which
it is associated. If one or more of the elements present have ionic radius and
valence similar to that of the trace element, the trace element can readily enter
a crystal structure as proxy for the abundant element. When the radii and charge
differ, substitution becomes more difficult and complex. A particularly interesting
set of trace elements are the rare earths. For the most part these have the same
valence (3) but their ionic radii differ systematically. For more than a decade
it has been known that the rare earths are fractionated differently when magmas
are formed or are partially crystallized. But few attempts have been made at
calibrating trace element partitioning within the range of natural abundances.
Mysen has now developed a general method for measuring their fractionation.
In one application he studied the partitioning of samarium between olivine,
orthopyroxene, and a hydrous silicate liquid, using the number of beta-tracks
generated by samarium- 151 as a measure of the amount of samarium present
in the various phases. By this means he was able to determine accurately the
concentration of samarium at levels as low as parts-per-billion.
Mysen found that only approximately 1 ppm Sm can enter olivine and ortho-
pyroxene before the activity coefficients of samarium become dependent on the
samarium concentration. That is, the trace element concentration range of be-
REPORT OF THE PRESIDENT
15
havior according to Henry's law is limited; the range is 25% larger in ortho-
pyroxene than in olivine. This concentration range in both minerals increases
about 15% per 100°C. He also found the crystal-liquid partition coefficients to
increase with pressure ; yet increasing temperature results in increased partitioning
of samarium into the olivine relative to liquid, whereas the partitioning decreased
in orthopyroxene relative to liquid. These results are already leading to a more
definitive understanding of what constitutes trace element behavior. The deter-
mination of partition coefficients between crystals and between crystals and
liquid under known conditions will provide a novel method, presumably inde-
pendent of major components, for ascertaining the conditions of formation of
igneous rocks.
DEPARTMENT OF TERRESTRIAL MAGNETISM
During the past decade we have learned a great deal about the origin and
evolution of the earth. We are now reasonably certain that the solar system was
formed 4.6 billion years ago. The oldest rocks that have been dated on earth are
about 3.7 billion years old. From that time to the present, rocks of different ages
were formed, attesting to a restless, changing earth. What of the gap between
4.6 b.y. and 3.7 b.y.? If the earth were our only source of specimens, we could
not hope to learn much about the hiatus. But space exploration has brought
information from other bodies such as meteorites, Mercury, Mars, and especially
the moon. As a result of a detailed examination of lunar samples, events on the
moon from 4.6 b.y. to 3.7 b.y. ago are now fairly well known. The moon, though
a comparatively small body, was at least partially molten very early. It is
generally agreed that the necessary heat was derived from gravitational impacts.
A large body such as the neighboring earth must have had a similar history,
with early melting. The lunar samples also provide information about large-scale
impacts that occurred about 4.0 b.y. ago. Thus there is now a reservoir of infor-
mation about the solar system that merits study and interpretation. Wetherill
has devoted considerable attention to this matter and has arrived at a model
for the accretional history of the solar system that is somewhat different from
the currently accepted view.
It is commonly believed that a very rapid, 10 3 to 10 4 year time-scale is a rather
firm constraint on the formation of the moon, and by inference, of the terrestrial
planets as well. The principal observational basis for this is the evidence for very
early chemical differentiation of the moon, coupled with thermal history calcula-
tions that purport to show that this requires rapid accretion to prevent the
limited gravitational energy of accretion from being wasted by radiation into
space. If true, such a constraint would be very useful in limiting the range of
permissible speculation. If false, it would be an obstacle to understanding what
really happened.
The problem is that all serious quantitative theories of planetary accretion
lead to time-scales of approximately 10 8 years, a discrepancy of 4 or 5 orders of
magnitude (the principal contributors in this area are V. S. Safronov and his
colleagues at the Institute for the Physics of the Earth in Moscow). Safronov
and Ruskol (his wife) have pointed out that the presence of large (about 100
km) accreting bodies could overcome the thermal problem if the heat were buried
below the large craters resulting from their impact. Unfortunately, a quantitative
application of Safronov's theory leads to an initially cool moon.
During the past year Wetherill has pursued this problem, using a stochastic
16 CARNEGIE INSTITUTION
j
simulation of the solar system based on earlier work by Opik and Arnold. This
leads to much larger bodies and higher velocities than those found by Safronov,
and consequently much greater accretional energy — now primarily the kinetic
energy of heliocentric motion of the bodies rather than the gravitational energy
resulting from falling into the gravitational "potential well" of the moon. These
differences arise from the "lumpiness" of the planetesimal distribution and
particularly the presence of the biggest lump, the embryo of Venus accreting
nearby. From this point of view, the initial temperature of the moon and planets
would be high on any accretional time scale, and the range of permissible thermal
and petrological models greatly increased.
Since the beginning, much of the research of the Department of Terrestrial
Magnetism has been strongly oriented toward global problems of the earth and
its history. Today a major program at the department is the study of the earth's
mantle, using the most advanced techniques of geochemistry and geophysics.
In the first few years of the "plate tectonic revolution" it seemed likely that
continents were a superficial feature of the earth, riding along on lithospheric
plates similar to those in oceanic regions. This view has been changing, as a
consequence of several developments: (1) Surface wave dispersion down to the
wavelength-limited depth of 200 km showed the absence of a low velocity zone
under South America. (2) Sacks and Okada have shown that beneath continents
Q and seismicity are high down to 350 km but decrease below that depth. (3) .
Isotopic and geochemical data on mantle-derived continental rocks show that
heterogeneities are preserved on a time-scale long compared to that of rates of
continental drift.
Because of these developments it is now believed that continental lithospheres
are fundamentally different from those below oceans, and some workers doubt
that there is any asthenosphere at all beneath continents.
This year Sacks and Snoke report direct evidence for a lithosphere-asthenosphere
boundary under South America at a depth of 400 ± 30 km associated with a
transition zone only a few kilometers thick. This evidence is in the form of a
seismic arrival that comes in between the p and s arrivals, which is interpreted
as resulting from an incident s phase converted to compressional waves at a
boundary. Unlike other models, this one fits all aspects of the data, such as
amplitude ratios of the different components. First-motion studies show the
polarity of the boundary — low velocity below, high velocity above. Thus the
bottom of a continent has been found, replacing the Moho, which has been known
to be superficial since acceptance of sea floor spreading and continental drift a
decade ago.
In some astrophysical work, Ford, Peterson, and Rubin report very high
resolution spectra of an e'liptical galaxy, spectra of such quality that at first
glance one would think they were those of a star rather than a galaxy. Previously,
high resolution spectra had been obtained for spirals in which the high abundance
of excited gas results in prominent emission lines. In contrast, these emission lines
are absent in the gas-poor elliptical galaxies; these new spectra are absorption
spectra, resulting from the integrated absorption of the stars in the galaxy. From
these spectra, velocity fields, a preliminary mass model, and a mass/light ratio
as a function of radius have been calculated. This development opens up for
observation the chemical and dynamical history of elliptical galaxies, a pre-
requisite for progress in understanding the reasons galaxies occur in two such
different forms. This development resulted from taking advantage of the high
spatial and spectral resolution of the 4-meter Carnegie image tube spectrograph
at Kitt Peak Observatory. The spectrograph is also being used to study absorp-
REPORT OF THE PRESIDENT 17
tion spectra of other galactic features, such as the stars in the central bar of
barred spiral galaxies.
HALE OBSERVATORIES
Earlier in this report I discussed the approaching completion of the Las
Campanas Irenee du Pont telescope and described some of its novel features.
Obviously, design and construction of such an instrument places heavy burdens
on staff, on some more than others. But the new construction in Chile was only
one of the tasks to be addressed. The other Hale telescopes must be continually
mantained and upgraded. Electronic control systems have evolved. In addition,
the use of electronics at the observing end of the telescope continues to increase,
and Hale staff are necessarily involved in the design, development, testing, and
use of such equipment. Added to all this are time-consuming tasks such as allo-
cation of observing time and dealing with 60 or more guest investigators.
In spite of all these burdens, which are borne by a relatively small staff, the
Hale Observatories are remarkably productive of advances in astronomy.
The new electronic devices create opportunities, and these are being exploited.
A major new frontier has been created by instruments that can count individual
photons. Sky background can be subtracted so that very distant, faint objects
that had been inaccessible can now be studied.
There are now available digital or linear photoelectric devices capable of
precise photometry on large numbers of picture elements within an optical image.
Such devices are equally important in the focal plane of a telescope and at the
output end of a spectrograph that is attached to the telescope. But the photo-
graphic plate, especially with the most modern emulsions of high information
content and optimum color sensitivity, continues to be improved and is in-
dispensable, particularly for wide-angle recording.
At the Hale Observatories, Sandage and Kristian, using the photographic
method, and Gunn and Oke, using an image tube, are reporting the discovery of
impressive numbers of new faint clusters of galaxies. Such clusters are important
because the brightest galaxy in a cluster has a standard luminosity very nearly
the same for any cluster. From this information the relative distances to such
clusters can be derived, and if the redshift of several galaxies in each cluster can
be measured with the spectrograph, new points in the Hubble diagram of redshift
versus distance can be plotted.
Not only are dozens of distant clusters of galaxies being found, but the pro-
duction of spectra and the measurement of redshifts of galaxies in them is
proceeding at a record rate. For this work Sandage, Kristian, and Westphal are
using an SIT (silicon intensifier target) Vidicon on a prism spectrograph at the
5-meter telescope, while Gunn and Oke are using the SIT- Vidicon with a grating
spectrograph.
Most of the new redshifts being measured are greater than z = 0.2 and a
substantial fraction have z > 0.4. Each of these galaxies is of small angular
extent, and most are much fainter than the uniform light of the sky (airglow
and zodiacal light) that overlies them. Spectroscopy involves subtracting elec-
tronically the spectrum of the sky from that of the object plus sky. The principal
limitation now is not measuring the spectra but rinding faint candidates and
positioning them on the slit. Observing depends upon low-level closed-circuit TV
systems for these operations, which formerly involved the much more difficult
technique of making blind offsets from nearby stars.
18 CARNEGIE INSTITUTION
Sandage has made an important discovery of substantial quantities of dust in
regions lying above the galactic plane. Though the finding itself was serendipitous,
it illustrates the benefits of intensive, long-term observational programs of the
kind the Hale Observatories can sponsor. During the last five years while looking
for remote clusters of galaxies, Sandage surveyed extensively the sky in the
region of the north and south polar gaps. During that survey he noted occasional
high latitude fields that have bright nebulous regions often in patterns correlated
over distances of degrees. He was able to show that these structures are reflection
nebulae whose illumination comes from our galactic plane.
Schmidt has analyzed spectroscopic and photometric observations of a com-
plete sample of quasi-stellar sources in the 6-cm National Radio Astronomy
Observatory survey. The value of V/V max , which measures the position of an
individual source within the available volume, appears to be correlated with the
radio spectral index. Sources with steep radio spectra have a mean V/V max
around 0.7, similar to values found previously for quasars in the 3C catalogue of
radio sources and the revised 4C catalogue, most of which have steep radio
spectra. This implies that the space density of these quasars rises steeply with
redshift (declines with cosmic time) by a factor of around 1000 to a redshift of
2.5. In contrast, quasars with flat or inverted radio spectra have a lower mean
value of V/Vmax around 0.52 ± 0.05 (m.e.), consistent with a uniform distri-
bution in space.
Searle is making a comparative study of the chemical history of the globular
cluster systems of the Galaxy, of M31, and the Magellanic Clouds. A more
sensitive method of determining the metal abundance in individual red giants
belonging to a cluster has been developed. In this report year Searle and Zinn
calibrated the new system and began a program to determine the metal abundance
of all globular clusters for which color-magnitude arrays are available. They
are especially interested in remote clusters and other stellar systems on the fringes
of the Galaxy.
In a related investigation Searle has obtained low resolution spectrophoto-
metric scans of 70 of the globular clusters in M31. So far all the observations
appear to be consistent with the idea that the M31 globular clusters differ from
one another in one parameter only — presumably the metal abundance. However,
there are differences between the globular clusters of M31 and the Galaxy, and
those of the Magellanic Clouds. There are also important differences between the
clusters of the Small and Large Clouds.
DEPARTMENT OF PLANT BIOLOGY
In the past ten years very powerful techniques have been developed to study
the structure and action of DNA. Substantial contributions to this important
effort have been made by Carnegie staff. The work of Britten and others at DTM
and of Brown and others at Embryology has been outstanding. Their work has
largely been devoted to animal DNA.
These techniques are now being applied to plants by Thompson and colleagues
at Plant Biology. Thompson reports this year on an analysis of the sequence
organization of pea DNA. The pea genome has a pattern much like that of the
toad Xenopus, with single-copy DNA sequences less than 3000 nucleotides long
interspersed with repetitive sequences. It is already clear that the study of plant
DNA will be a powerful tool in the elucidation of evolutionary relationships in
plants. A typical question is the origin of C 4 photosynthesis in Atriplex (saltbush)
and the DNA sequence comparison between C 3 and C 4 species of that genus.
REPORT OF THE PRESIDENT 19
Using techniques available for examining DNA, Rogler, a fellow, and Thomp-
son have made a useful contribution to knowledge of crown gall tumors and to
the controversial topic of recombinant DNA.
Wounded plant tissue infected with Agrobacterium tumefaciens will, under
appropriate conditions, form tumors that have several physiological properties
in common with mammalian cancer. These properties include non-self-limiting
growth on synthetic medium devoid of hormones and the inability to differentiate
into organized, functional tissue in response to treatments causing such differenti-
ation in normal cultured cell lines. Crown gall tumor cells retain these properties
when cultured in the absence of the inciting organism and, when grafted into
healthy plants, produce tumorous overgrowths.
Recently it was discovered that all pathogenic strains of Agrobacterium
tumefaciens contain one or more large plasmids 1 and that loss of the plasmid
results in loss of virulence. 2 On the other hand, transformation with plasmid
DNA results in acquisition of virulence by the strain acquiring the plasmid.
These two lines of evidence strongly support the hypothesis that the plasmid is a
necessary genetic element for virulence in Agrobacterium.
Rogler and Thompson have described changes in plasmid DNA sequences in
Agrobacterium during either the loss or the acquisition of virulence. In their
report they discuss an instance in which a large deletion in the plasmid genome
is associated with loss of virulence. They also present evidence that acquisition
of virulence may involve replacement of a preexisting avirulent plasmid with a
virulent plasmid derived from another strain of Agrobacterium.
Rogler and Thompson do not attempt to relate their findings to the current
controversy over DNA work — perhaps wisely so. One can choose to view the
results with alarm or with equanimity. Today much recombinant DNA work
uses plasmids of Escherichia coli. The fact that a change in a plasmid of another
bacterium gives rise to crown gall tumors argues for caution when producing
new kinds of plasmids in Escherichia coli. A contrary view is that the patho-
genicity of Agrobacterium tumefaciens arose naturally and doubtless is the
result of innumerable experiments performed by nature. It may well be true, as
some argue, that among the countless experiments on DNA performed by nature
everything and every possible combination has already been tested.
The physiological ecology group continues to make good progress in under-
standing the factors that enable plants to cope with stress and thus to exist under
extreme conditions of temperature, moisture, light intensity, and other variables.
The general approach is to combine field observations using mobile equipment
with laboratory studies. When plants having unusual characteristics are noted
in nature, their photosynthetic machinery can be studied in the laboratory.
In the current report, Mooney, Bjorkman, Ehleringer, and Berry describe
some of their efforts in studying the native flora of Death Valley, California.
Although Death Valley is often identified as one of the hottest and driest en-
vironments on earth, these extremely high air temperatures (exceeding 50°) occur
only in the summer. In winter, relatively cool temperatures (20°C) predominate,
and in spring and fall air temperatures are often around 30° C. Consequently,
1 Zaenen, I., N. van Larebeke, H. Teuchy, M. van Montagu, and J. Schell, J. Mol. Biol, 85,
109-127, 1974.
2 van Larebeke, N., O. Engler, M. Holsters, S. van den Elsaker, I. Zaenen, R. A. Schilperoort,
and J. Schell, Nature, 225, 169-170, 1974.
Watson, B., T. C. Currier, M. P. Gordon, M. D. Chilton, and E. W. Nester, J. Bacteriol.,
123, 255-264, 1975.
20 CARNEGIE INSTITUTION
Death Valley provides a broad variety of thermal environments and a unique
opportunity to study the ways plants adapt to these seasonal thermal regimes.
Utilizing the mobile laboratory, the photosynthetic capacity of plants was
measured during the winter (January), spring (March), summer (late May),
and fall (October) seasons. Plants studied included a winter annual (Camissonia
claviformis) , a summer active herbaceous perennial (Tidestromia oblongifolia) ,
and two evergreen shrubs (Larrea divaricata and Atriplex hymenelytra) .
Under natural conditions the two evergreen shrub species were photosyntheti-
cally active throughout the year, although their maximum photosynthetic rates
were considerably below those of the two that hold their leaves for a short period.
The two herbaceous species, Tidestromia (C 4 ) and Camissonia (C 3 ), have ex-
tremely high maximum photosynthetic rates; yet their physiological activities
are restricted for the most part to a single season.
The Cs winter-active annual Camissonia germinates following heavy winter
rains and can, under certain conditions, complete its entire life cycle in six
weeks. This short-lived species has a remarkable capacity to capture sunlight
and fix carbon. Camissonia converts incident photosynthetically active radiation
(400-700 nra) to chemical energy with an efficiency of 8.5%. This is possible
because of the lack of light saturation at midday irradiances, a feature that is
also characteristic of Tidestromia. The in situ midday photosynthetic rate of
Camissonia was higher than the rate measured for such productive crop species
as corn, sorghum, and sugar cane.
Earlier studies had identified Tidestromia oblongifolia as a plant with a high
rate of photosynthesis at elevated temperatures. During the year Bjorkman,
Boynton, and Berry participated in studies to determine the thermal stability of
key components of the photosynthetic machinery responsible for the marked
differences in temperature tolerances between thermophilic higher plant species
such as T. oblongifolia and cool-temperate plants such as Atriplex sabulosa. As
reported last year, the two C 4 species differ in their high temperature stability
in several respects, including the maintenance of semipermeability of the cell
membranes, respiratory activity, and photosynthetic activity. While onset of high-
temperature damage differed by about 10°C in the two plants, photosynthesis
was affected at considerably lower temperatures than either respiration or mem-
brane semipermeability.
This year they studied the effect of temperature on the quantum yield for
photosynthesis by intact leaves with its effect on photosystem I and II activities;
ribulose 1,5-diphosphate carboxylase activity (the C 3 enzyme for C0 2 fixation) ;
phosphoenolpyruvate carboxylase activity (the C 4 enzyme for C0 2 fixation) ; and
heat coagulability of soluble proteins as a measure of irreversible denaturation.
It is clear that the best fit with thermal inhibition of photosynthesis in both
cases is with system II activity. In addition, however, there were interesting
differences in heat stability of the C 3 enzyme from the two plants, suggesting that
not just lipids but proteins may differ between plants adapted to high tempera-
tures and those adapted to low temperatures.
DEPARTMENT OF EMBRYOLOGY
James D. Ebert, the retiring director, leaves this department after having
brought it successfully through a major era of evolution from classical embryology
to a new form of developmental biology heavily based on molecular biology.
The two major classes of studies at this department are (1) broad-ranging
REPORT OF THE PRESIDENT 21
fundamental research on DNA — its structure in chromosomes and its behavior
as genetic material and (2) researches aimed at better understanding of the
structure, behavior, and interactions of membranes. The investigations of DNA
are progressing very well and the department has compiled an impressive record
of achievements in this field. The studies of membranes are a newer, evolving
activity in the department which have already been fruitful and which have
great potential. The staff investigators (Fambrough, Muller, and Pagano) are
young and have been brought together fairly recently. Of the three, Fambrough
has been at the department longest and has accumulated an impressive record
of accomplishment in studies of nerve-muscle interactions with emphasis on the
acetylcholine (ACh) receptors in skeletal muscle fiber membranes. In a series
of articles Fambrough and his students and colleagues report on the appearance
of ACh receptors during muscle development, the production and incorporation
of ACh receptors into the surface membranes of myotubes and their subsequent
degradation, the clustering of ACh receptors at newly formed neuromuscular
junctions both in vitro and in vivo, the number and distribution of ACh re-
ceptors in adult rat skeletal muscle fibers, and the time course of change in these
parameters following denervation. They are continuing to explore physiological
aspects of receptor function, examining further the biosynthesis and degradation
of ACh receptors and obtaining more accurate estimates of ACh receptor density
on denervated muscle fiber membranes.
In this year's report, Fambrough and his student Peter Devreotes present a
model for the major events in receptor metabolism. The protein subunits that
are to form the ACh receptor are synthesized inside the muscle cells and are
rapidly assembled into receptor units in the cytoplasm of the muscle cells where
they constitute a "precursor pool." The average residence of a new receptor in
the precursor pool is 2 to 3 hours. Receptors are transferred from this pool to
the surface by an energy-requiring process and are then able to function. The
average half-life of an ACh receptor in the plasma membrane is 22 hours. Re-
ceptors are degraded by an energy-requiring process that involves the internali-
zation of the receptor and transport to secondary lysosomes and then proteolytic
destruction.
In his continuing professional development, Pagano illustrates one of the
principles on which the Institution is based: Andrew Carnegie's advice that the
Institution seek out talented individuals and support them.
Eber-t followed this course in choosing and fostering Pagano. Four years ago,
when this young scholar came to the department, his background was that of a
physical chemist who had worked with some nonliving model membrane systems.
He has since acquired substantial experience in cellular biology, and he and his
colleagues have been using artificially generated lipid vesicles to perturb mem-
brane components in living cells. In the report of the director they present findings
on the interactions of phosphatidyl choline vesicles with murine lymphocytes,
and the physiological consequences of these interactions. Pagano and postdoctoral
fellows Huang and Ozato have shown that the pathways of vesicle-cell inter-
actions enumerated in previous reports (vesicle-cell fusion, adsorption, and lipid
exchange) are exquisitely sensitive to the molecular composition and physical
properties of the artificial vesicles. In the lymphocyte system, the resulting
membrane perturbations lead to altered surface properties and greater prolifera-
tive activity of these cells when stimulated by various mitogens and antigens.
Muller, the most recently added staff member, is tackling the major question
of how single nerve cells are organized into a nervous system. He is especially
looking at the pathways neurons take to reach their targets. For this purpose he
22 CARNEGIE INSTITUTION
has been studying activities in the nervous system of the leech. This creature
has a very simple nervous system, yet its neurons behave in many ways like
the neurons of higher animals.
Neurons in this segmented worm are grouped 350 together to form essentially
identical ganglia that are linked in a chain by axon bundles; each ganglion
mediates the reflexes and controls the behavior of its segment. It has been
possible to monitor electrical activity from within single sensory and motor
neurons with microelectrodes and to correlate the physiological activity of func-
tionally identified neurons with their characteristic shape and position within
each ganglion. By recording from pairs of neurons, one can trace synaptic con-
nections, which have been found to be surprisingly uniform and in that sense
predictable between specific neurons.
In his current report, Muller describes the structural details of chemical
synapses, delineates their arrangement on sensory neurons, and explores the
sprouting of sensory neuron terminals.
Staff members active in DNA research include Brown, Dawid, Reeder, and
Suzuki. In a series of investigations inspired by Brown and Dawid, this group
together with fellows and guest investigators have made a significant impact in
developmental, molecular, and evolutionary biology. Studies of "developmental
genetics by gene isolation" have led to new and brilliantly successful approaches
to fundamental questions in development. The department has originated and
exploited techniques for gene purification. Studies of "gene anatomy" — the struc-
ture and arrangement of genes — have resulted in new insights into the mecha-
nism by which multiple tandem genes evolve together.
Brown has a long-term goal of reassembling the macromolecules that enable
DNA to be transcribed with fidelity in the living cell. In this way he hopes to
elucidate the control mechanisms of animal genes. For this purpose he and his
colleagues are now concentrating on the family of genes from Xenopus which
code for the small RNA molecule of ribosomes termed 5S RNA.
This year Brown's group reports progress along several lines. The "transcrip-
tion unit," that is, the region of the Xenopus DNA that is transcribed into 5S
RNA in the living cell, is relatively simple — only 120 base pairs long. Moreover,
the form of RNA polymerase that transcribes this gene in vivo has been identi-
fied. The fact that the mature RNA transcript does not appear to be derived from
a higher molecular weight precursor is especially important: It means that the
true initiation site of transcription from 5S DNA is known. Therefore, one can
detect faithful transcription when it occurs in vitro.
Three different kinds of 5S DNA have been cloned in E. coli by means of the
new recombinant DNA methology, thereby providing an order of magnitude
more material to work with and permitting the careful examination of a single
repeating unit of a tandem gene family. The "restriction maps" of these cloned
DNAs — the sites where they are cleaved by certain restriction enzymes — are
described by fellows Jeffrey Doering and Scott Emmons. Emmons is attempting
to develop an assay for proper transcription. Fellow Nina Federoff is exploring
DNA sequencing methods to determine the exact nucleotide sequence in the
spacer regions of 5S DNA, particularly those regions adjacent to the initiation
site of the gene.
DNA structure continues to hold the attention of several staff members and
fellows. Reeder and Dawid, along with fellow Peter Botchan, report on the
structure of nontranscribed spacer regions in Xenopus ribosomal DNA, based on
studies using cloned fragments of rDNA generated with the restriction endo-
nuclease EcoRI. Fellow Peter Wellauer and Dawid, a staff member, have under-
REPORT OF THE PRESIDENT
taken the first stages of a study of the ribosomal DNA of Drosophila; Dawid
and his colleagues are cloning Xenopus mitochondrial DNA; and Carnegie fellow
Ohshima and staff member Suzuki are attempting to clone the silk fibroin gene.
In all these studies, the department has adhered to the safety standards pro-
posed by the Asilomar Conference on Recombinant DNA.
EDUCATION
To an unprecedented extent this country has become a nation of the here-and-
now with a great part of its efforts and resources devoted to instant gratification.
One of the few but important investments we make for the future is in education.
For most people, formal education stops at graduation from college or before;
comparatively few of these go on for advanced degrees in the sciences to form
the cadre from which our future leaders in science and technology will emerge.
But of the cadre only a fraction achieve creativity. After twelve years of primary
and secondary schools and four years of university, most young people have
mastered the memorizing of facts, have developed some judgment and some
synthetic abilities, and have arrived at a personal identity. But the effort needed
to become a creative scientist and to establish a professional identity is far more
demanding than the struggles of adolescence. Among other things, what is in-
volved is a metamorphism from a fact-accumulating to an idea-producing
organism — a change from depending on the direction of others to confidently
relying on one's own judgments and initiatives. The last two predoctoral years
and the first three postdoctoral years are the critical period. During that time
only those few who achieve the metamorphism are launched on a productive
career. There is evidence to show that a crucial factor in the outcome is the
intellectual environment in which the young scholar is immersed. At the Carnegie
Institution we have diligently sought, and I believe we have succeeded in pro-
viding, optimum circumstances for the metamorphism of our fellows. In turn we
feel that our fellowship program is one of the best kinds of investment that the
Institution and the nation makes for the future.
Each year the departments of the Institution makes contributions to knowledge
that win for them continuing recognition as leaders in their fields. At the same
time staff members enjoy a high reputation for innovation. This aura of success
attracts young scholars, many of whom seek to associate with experts in their
chosen fields and to savor the excitement and intellectual growth that a creative
group fosters. Since the departments can choose from among many excellent
candidates, the quality of our predoctoral and postdoctoral fellows is excellent.
They merit careful tuition and fostering, and they get them.
An important factor is the Institution's policies with respect to staff and
fellows. Staff members are expected to participate actively and personally in
research and teaching. They do not sit at command posts in offices while others
work for them. They devote almost all their time to scholarly activities with
few interruptions for administrative activities. Thus they are highly available
for discussions. Since the ratio of staff to fellows is about one to one, the young
scholars can have all the consultation and instruction they want. The situation
is a modern version of the ideal enunciated nearly a century ago of professor and
student sitting on a log discoursing together.
Optimal intellectual growth of fellows is promoted by another policy of the
Institution. They are encouraged to pursue projects of their own choosing, and
24- CARNEGIE INSTITUTION
they are given the kind of back-up in equipment and supplies that is enjoyed by
staff members.
The precise patterns of training differ from department to department, reflect-
ing the needs and opportunities of the respective fields of inquiry. However, some
general patterns prevail throughout the Institution, and these are outlined and
illustrated in an essay on education in the Department of Embryology that was
prepared for this report by the retiring director, James D. Ebert. His comments
follow:
For the past two decades training has been an integral part of the
department's program. It was clear in 1956, as the department prepared
to embark in new directions in molecular developmental biology and
genetics, that even if ample funds and space were available immedi-
ately to undertake all the tasks that had been set, the scientific talent
was not. The world's pool of able developmental biologists then was
small. President Haskins and the director agreed that the catalysis
provided by young minds — graduate students and postdoctoral fellows —
would be a vital ingredient in the shaping of these new fields.
The wisdom of that decision has been demonstrated repeatedly. A
glance at today's staff tells a part — but only a small part — of the story.
Director-designate Donald Brown and staff members Igor Dawid,
Douglas Fambrough, Ronald Reeder, and Yoshiaki Suzuki all joined the
department initially for training. Thus, students become teachers and
they, in turn, set standards of excellence for new generations of students
to follow.
Douglas Caston, Professor and co-Director of the Developmental
Biology Center at Case Western Reserve University, emphasized this
point recently when he wrote "The experience I gained as a fellow [at
Carnegie] has been the major force in shaping not only my personal
career but in my helping junior colleagues shape their careers."
The department's purpose in offering its training program is to
encourage the development of research scientists. First consideration
is given to students who appear likely, by their intellectual gifts and
intentions, to contribute to the advancement of the field whether in
biology or in medicine. The previous training of applicants, whether in
embryology, molecular biology or genetics, pediatrics or physics, may be
of secondary importance, preference being given to the most promising
students. Thus the department's program has had a profound impact
not only in molecular developmental biology but in the clinical and
mission-oriented fields that draw from the same spring. Caston is a
molecular development biologist, yet his experience is matched by that
of Glenn Rosenquist, one of the nation's leading students of cardiology
and birth defects, now Professor of Pediatrics at the University of
Nebraska Medical Center, who wrote "The two years I spent at Car-
negie Institution were probably the most important years of my entire
career."
Although over the years opportunities have been provided for the
training of undergraduate students of exceptional promise, most of the
department's teaching has been focused on graduate students and
postdoctoral fellows. It is at the graduate and postgraduate levels that
the department's unique resources — both human and scientific — can
have their greatest impact. Several generations of fellows have written
about the department's unique intellectual climate. The department
provides well-organized and well-equipped laboratories where, in Dorothea
Rudnick's words, "Everything peripheral [is] taken care of." In such
an environment the most promising young people can be afforded the
REPORT OF THE PRESIDENT
opportunity to exercise and extend their abilities, enabling them to take
their critical steps toward independence.
A listing of the department's graduate students and postdoctoral
fellows constitutes a veritable "Who's Who" of developmental biology.
Even a cursory glance reveals Carnegie fellows teaching and carrying
forward their research at Harvard, Princeton, Brown, and Rockefeller
Universities; at Virginia, Emory, Case Western Reserve, and Washing-
ton University, St. Louis; at the Universities of California (Berkeley),
Kyoto, Tokyo, and London. They are working actively in departments
of biology, in schools of medicine, and in an enormous variety of both
basic and clinical research institutions.
Thus far we have emphasized the "products" of training, that is, the
students themselves. Yet there is another factor to be considered: The
greatness of the department's contribution to science can itself be
attributed in part to its insistence that teaching and research are
inextricably interwoven. We spoke earlier of the catalysis provided by
young minds. The truth of that observation is reiterated year after
year in the department's Annual Reports. The interaction between staff
member and student is truly reciprocal, with students being an intimate
part of some of the department's greatest achievements. The classic
demonstration of the role of the nucleolus in the synthesis of ribosomal
RNA was provided by a staff member, Brown, working with a fellow,
Gurdon. Similarly, two fellows, Yoshiaki Suzuki and Patrick Gage,
worked closely with Brown in the isolation and characterization of the
gene for silk fibroin. Fambrough's far-reaching studies of the acetyl-
choline receptor sites in muscle have been advanced through the con-
tributions of graduate students Hartzell and Devreotes. Fellow Peter
Wellauer, working with Brown, Dawid, and Reeder, has led the way
in elucidating gene structure.
The crucial nature of the intellectual climate needed for the training
of young investigators is emphasized in our earlier expression "steps
toward independence." Each student or fellow must be considered an
individual. It is difficult, some may say impossible, to teach another
individual how to do research except by example. There must be a
special interplay between student and preceptor. The student or fellow
must be afforded an ample opportunity to, in the words of former
fellow Malcolm Steinberg, now Professor of Biology at Princeton Uni-
versity, "follow his own nose." Steinberg went on to say "The trail I
picked up as a fellow in the Department of Embryology has broadened
' into a significant field of investigation now pursued by many scientists
the world over." Turning again to the clinical side, Gilbert Greenwald,
Professor of Obstetrics and Gynecology at the University of Kansas
Medical Center, wrote "The freedom that I had to pursue a number
of my own ideas gave me the confidence to think that I could embark
on my own career as an independent and . . . imaginative investigator.
I believe the last point is the most important objective to be gained from
any postdoctoral experience.
LOSSES . . .
I am sorry to report that William I. Myers died in January. For 27 years the
board of trustees benefited from the insight and knowledge he acquired during
a career of distinguished service in government and education.
As one of President Roosevelt's top agricultural aides from 1933 to 1938,
Dr. Myers established and headed the Farm Credit Administration. He also
26 CAENEGIE INSTITUTION
served as president of the Federal Farm Mortgage Corporation and the American
Farm Economics Association. In addition, he was director of the Federal Surplus
Relief Corporation and the Commodity Credit Corporation.
In 1938 Dr. Myers accepted the chairmanship of the Department of Agri-
cultural Economics at Cornell University. He became dean of the department
in 1943 and held that post until his retirement in 1959. He was director of the
Eisenhower Exchange Fellowships from 1953 to 1969 and was a trustee of the
Agricultural Development Council from 1954 to 1971. He was also a trustee of
the Rockefeller Foundation (1941-1957) and later served on its Board of Con-
sultants for Agricultural Sciences until 1965.
At home in New York State, he was part of the Science and Technology
Foundation and a member of the Council of Economic Advisers and a director
of the Association for Crippled Children and Adults of New York. Dr. Myers
was a director or trustee of six companies as well as Marine Midland Banks, Inc.,
and the New York State Gas and Electric Corporation.
Another loss to the Institution is Berwind P. Kaufmann, the last director of
the Department of Genetics (1960-1962), who died in September 1975. After
distinguishing himself as a professor at Southwestern College and at the Uni-
versity of Alabama, he joined Carnegie Institution in 1936 and remained with
us until our research in genetics was phased out in 1962.
At the department, Dr. Kaufmann's cytological and cytogenetic research made
important contributions to the field, among them the pioneering discovery that
chromosomes contain double spiral bands and the first demonstration of inter-
change between X and Y chromosomes. His studies indicated the association of
RNA with a histone-type protein and of DNA with a tryptophane-rich protein
many years before confirmatory biochemical data were obtained. He also studied
the effects of radiation, especially on chromosomes, and was a member of the
National Academy of Sciences' Committee on Genetic Effects of Atomic Radiation.
After retiring from the Institution, Dr. Kaufmann continued his genetics
research at the University of Michigan, Ann Arbor, as professor of zoology and
botany and senior research scientist. He was made professor emeritus when he
retired from the University in 1967.
I must also report with sorrow the death in January of Rudolph Minkowski, a
staff member of the Hale Observatories from 1937 to 1960. A native of Germany,
he fled to this country in 1935 and became an American citizen in 1940.
Throughout his career he collaborated with another German-born Observa-
tories' astronomer, Walter Baade. Together they studied supernovae, which they
divided into two types on the basis of long sequences of supernovae spectra first
obtained by Minkowski. With Baade, he became interested in radio astronomy,
and in the 1950s was the first to identify and interpret correctly a radio source,
the remnant of a long-dead supernova in the constellation of Cassiopeia. For
years Dr. Minkowski supervised the National Geographic Society-Palomar
Observatory Sky Survey.
Dr. Minkowski received the Bruce Medal of the Astronomical Society of the
Pacific in 1961 and an honorary doctorate from the University of California,
Berkeley, in 1968. After his retirement from Hale, he worked at the University
of California's radio astronomy laboratory.
The decision this year by an exceptionally gifted scientist and administrator
to leave the Institution is, inevitably, cause for regret. Yet our sense of loss is
REPORT OF THE PRESIDENT dl
assuaged by our pride in the accomplishments of James D. Ebert during the
twenty years he was director of the Department of Embryology. And we look
forward to Dr. Ebert's contributions to developmental biology at the Marine
Biological Laboratory, Woods Hole, Massachusetts, where he is the first full-time
president and director.
Since his appointment as director in 1955 by the late Vannevar Bush, Dr.
Ebert has guided the Embryology laboratory through a metamorphosis that
took it from a place in the forefront of classical embryology to a leading role
in molecular developmental genetics.
Among the outstanding achievements of the Department during Dr. Ebert's
administration are his own work on the cellular and molecular basis of heart
development in embryos, and his studies on the graft-versus-host reaction, the
relation of viruses to tumor formation, and the regulation of DNA synthesis.
Moreover, his leadership and encouragement created a stimulating and supportive
environment in which the research of his colleagues and students also flourished.
In addition to its standing as a foremost research center, the Department of
Embryology has become, under Dr. Ebert's guidance, internationally known for
its advanced training program in developmental biology.
Dr. Ebert's contributions to the scientific community go beyond his research
and his service to his "home" institution. He is professor of biology and pro-
fessor of embryology at The Johns Hopkins University, and he has lectured as
a visiting professor at other institutions in the United States and abroad. He is
the first chairman of the Assembly of Life Sciences of the National Academy of
Sciences; past president of the American Institute of Biological Sciences, the
American Society of Zoologists, and the Society for the Study of Development
and Growth. He is a trustee of the Board of Scientific Overseers of the Jackson
Laboratory, a member of the Corporation of Woods Hole Oceanographic Institu-
tion, and he is on the board of visitors of ten universities. He has served as a
consultant to the U.S. Department of Health, Education and Welfare. And he
has been the editor or a member of the editorial board of half a dozen biological
journals.
. . . AND GAINS
It is our privilege to welcome to the Institution this year two men of out-
standing achievement in their fields. Robert 0. Anderson and J. Paul Austin were
elected trustees at our annual meeting of the board on April 30.
Robert 0. Anderson, chairman of the board and chief executive officer of
Atlantic Richfield Company, has held those positions since 1965. His experience
in the business world is extensive and varied. He joined the American Mineral
Spirits Co. in his native Chicago soon after his graduation from the University
of Chicago in 1939. Within two years he had set up his own corporation, Malco
Refineries, Inc., which later became Hondo Oil and Gas Co. and merged with
Atlantic Richfield in 1963. At that time Mr. Anderson became one of the directors
of Atlantic Richfield.
Although the focus of Mr. Anderson's business career has been in the petroleum
industry, he has been involved in such diverse enterprises as mining, milling, and
general manufacturing. In addition, he operates his own livestock company in
Lincoln County, New Mexico. His experience is further broadened by being a
28 CARNEGIE INSTITUTION
trustee of three universities, and a director of the First Chicago Corporation and
the First National Bank of Chicago. His participation in cultural, civic, and
charitable organizations includes chairmanships with the Aspen Institute for
Humanistic Studies and The Lovelace Foundation.
J. Paul Austin, chairman of the board and chief executive officer of the Coca-
Cola Company, has enjoyed a distinguished career with that firm since 1949.
Joining the legal department of the Company in that year, Mr. Austin trans-
ferred to the Coca-Cola Export Corporation the following year and became
the Corporation's president in 1959. In 1962 he was elected the tenth president
of the Coca-Cola Company, in 1966, its chief executive officer, and in 1970,
chairman of the board.
Mr. Austin also holds directorships with other well-known companies, among
them Morgan Guaranty Trust Co. of New York; Dow Jones and Co., Inc.; and
the General Electric Co. He is a trustee of numerous organizations, including the
United States Council of International Chambers of Commerce, the National
Industrial Conference Board, the Rand Corporation, the Twentieth Century
Fund, and the California Institute of Technology. He is also a regent of the
Smithsonian Institution.
Mr. Austin was born in Georgia in 1915 and received his A.B. (1937) and
LL.B. (1950) degrees from Harvard University. He served with the U.S. Navy
in the Pacific during World War II, achieving the rank of lieutenant commander.
He was awarded the Legion of Merit.
When James Ebert moved to Woods Hole, we were especially fortunate to
have Donald D. Brown to take over the direction of the Department of Embryol-
ogy. Since 1961 when he was brought to the department as a promising fellow
by Ebert, Brown has matured in the Carnegie environment, becoming a pioneer
in developmental biology and molecular genetics and a gifted teacher.
Brown attended Dartmouth College and received the M.S. (in biochemistry)
and the M.D. degrees from the University of Chicago in 1956. After interning
at Charity Hospital, New Orleans, he embarked upon a research career, first as
a research associate in the National Institute of Mental Health (1957-1959)
and then as a special fellow of the National Cancer Institute in the Service de
Biochimie Cellulaire, Institut Pasteur, Paris. After a year as fellow at the
Department of Embryology, he was appointed staff member.
A leader in developing techniques for the recombination of animal genes with
bacterial DNA molecules, Dr. Brown was one of the first to isolate animal
genes — from the South African clawed toad Xenopus laevis. He has also studied
the fibroin gene of the larval silkworm Bombyx mori. In recognition of these
accomplishments he was chosen by the National Academy of Sciences to receive
the U.S. Steel Foundation Award in Molecular Biology (1973) and was awarded
the V. D. Mattia Lectureship of the Roche Institute of Molecular Biology (1975).
In 1976 he received the Distinguished Alumni Award from the University of
Chicago Medical School and an honorary D.Sc. from the University of Chicago.
A frequent contributor to scientific journals, Dr. Brown became editor of
Developmental Biology in 1975. He has been president and treasurer of the
Society for Developmental Biology and is a member of the National Academy of
Sciences and a fellow of the American Academy of Arts and Sciences. Con-
current with his work at Carnegie, Dr. Brown is part-time professor of biology
at The Johns Hopkins University.
REPORT OF THE PRESIDENT «B
The following honors have been awarded to Staff Members during the year:
L. Thomas Aldrich of the Department of Terrestrial Magnetism was re-elected
General Secretary of the American Geophysical Union.
Peter M. Bell of the Geophysical Laboratory received the Exceptional Scien-
tific Achievement Medal of the National Aeronautics and Space Administration.
Donald D. Brown of the Department of Embryology was honored by his
alma mater with the Distinguished Alumni Award of the University of Chicago
Medical School and a D.Sc. degree from the University of Chicago. He also
received the V. D. Mattia Lectureship of the Roche Institute of Molecular
Biology and gave the John M. Prather Lectures in Biology at Harvard University.
Jesse L. Greenstein of the Hale Observatories received several awards, includ-
ing the Distinguished Public Service Medal of the National Aeronautics and
Space Administration, a Medal of the University of Liege, and the Gold Medal
of the Royal Astronomical Society.
James E. Gunn, Hale Observatories, was the first George Ellery Hale Visiting
Professor at the University of Chicago.
Elburt F. Osborn was given the 29th Albert Victor Bleininger Award by the
Pittsburgh Section of the American Ceramics Society.
The Institution has been fortunate in the generosity of its friends and associates.
Gifts from staff, former staff, and trustees, as well as other individuals and
organizations, have added $1,909,000 to the funds for research over the past six
years. Some of the recent donors are the late Dr. Vannevar Bush, Mr. William
Golden, Mr. and Mrs. Crawford H. Greenewalt, Mr. William McChesney Martin,
Jr., Mr. John C. Meyer, Mr. Henry S. Morgan, Mr. Walter H. Page, Dr. George
W. Preston, and Mr. Orley H. Truman.
30 CARNEGIE INSTITUTION
FACULTY, FELLOWS, AND STUDENTS
1975-1976
DEPARTMENT OF EMBRYOLOGY
Baltimore, Maryland
Director
James D. Ebert
Director Designate
Donald D. Brown
Staff Members
Igor B. Dawid
Douglas M. Fambrough
Kenneth J. Muller
Richard E. Pagano
Ronald H. Reeder
Yoshiaki Suzuki
Fellows
Peter Botchan
Diana Card
S. T. Carbonetto
Jeffrey Doering
Scott Emmons
Nina Fedoroff
Paul Geshelin
Elizabeth Godwin
Leaf Huang
Yasumi Ohshima
Keiko Ozato
Aileen K. Ritchie
Alex Sandra
Masatoshi Takeichi
Katherine Tepperman
Harvey Wahn
Peter K. Wellauer
Students
Peter Devreotes
John Gardner
Carol Kaushagen
Jose Ramirez
GEOPHYSICAL LABORATORY
Washington, D.C.
Director
Hatten S. Yoder, Jr.
Carnegie Institution Distinguished Professor
Elburt F. Osborn
REPORT OF THE PRESIDENT SI
Emeritus
Emanuel G. Zies
Staff Members
Peter M. Bell
Francis R. Boyd, Jr.
Felix Chayes
John R. Cronin
Gordon L. Davis
David H. Eggler
Larry W. Finger
John D. Frantz
P. Edgar Hare
Thomas C. Hoering
T. Neil Irvine
Thomas E. Krogh
Ikuo Kushiro
Ho-Kwang Mao
Yoshikazu Ohashi
Douglas Rumble III
David Virgo
Fellows
Richard J. Arculus
Nicholas T. Arndt
Timothy M. Benjamin
John M. Ferry
Anthony A. Finnerty
Toshitsugu Fujii
John I. Hedges
Frank E. Huggins
Gifford H. Miller
Bj0rn 0. Mysen
Howard R. Naslund
Frank S. Spear
Juergen Trochimczyk
Rosemary J. Vidale
E. Bruce Watson
Bernard J. Wood
Students
Julia A. Dill
Dora Y. Lee
Catherine A. McCammon
HALE OBSERVATORIES
Pasadena, California
Director
Horace W. Babcock
Associate Director
J. Beverley Oke
CARNEGIE INSTITUTION
Staff Members
Halton C. Arp
Eric E. Becklin
Edwin W. Dennison
Jesse L. Greenstein
James E. Gunn
Robert F. Howard
Jerome Kristian
Robert B. Leighton
Guido Munch
Gerry Neugebauer
S. Eric Persson
George W. Preston
Bruce H. Rule
Allan R. Sandage
Wallace L. W. Sargent
Maarten Schmidt
Leonard T. Searle
Stephen A. Schectman
Arthur H. Vaughan, Jr.
James A. Westphal
Harold Zirin
Staff Associates
Robert J. Brucato
Michael W. Werner
Fellows
A. Ger de Bruyn
Eduardo Hardy
Mark Hartoog
Francois Schweizer
Christopher P. Wilson
Hirokazu Yoshimura
Robert J. Zinn
Carnegie-Chilean Fellow
Maria Teresa Ruiz
Student Observers
Steven V. W. Beckwith
Kirk Borne
Todd Boroson
France Cordova
David J. Diner
Jonathan H. Elias
Richard F. Green
Paul Hickson
John G. Hoessel
John P. Huchra
Steven Kent
John Kormendy
Barry J. Labonte
Philip Massey
Jorge Melnick
Daniel Nadeau
REPORT OF THE PRESIDENT S3
William C. Priedhorsky
Douglas M. Rabin
Russell 0. Redman
Anneila I. Sargent
William L. Sebok
David Sholle
Richard J. Terrile
Edwin L. Turner
Barry E. Turnrose
Richard Wade
Theodore B. Williams
Steven P. Willner
DEPARTMENT OF PLANT BIOLOGY
Stanford, California
Director
Winslow R. Briggs
Staff Members
Joseph A. Berry
Olle Bjorkman
Jeanette S. Brown
David C. Fork
Malcolm A. Nobs
William F. Thompson
Emeritus
C. Stacy French
William M. Hiesey
Fellows
Michael Blatt
Steven J. Britz
James Collatz
John Cross
Ulrike Pohrmann
James Ehleringer
Algirdas J. Jesaitis
George Johnson
Aaron Kaplan
John Mackenzie
Bruce E. Mahall
Norio Murata
Michael Murray
Charles E. Rogler
Ulrich Schreiber
Diana Stein
Alan Stemler
Heather Strong Belford
Students
Brian Anderson
Robert D. Brain
Daniel Cosgrove
34
CARNEGIE INSTITUTION
William Curtis
Mary Enama
Thomas Payne
Richard Preisler
Karen Swift
DEPARTMENT OF
TERRESTRIAL MAGNETISM
Washington, D.C.
Director
George W. Wetherill
Distinguished Service Member
Merle A. Tuve
Emeritus
Scott E. Forbush
Staff Members
L. Thomas Aldrich
George E. Assousa
Louis Brown
Dean B. Cowie
W. Kent Ford, Jr.
Stanley R. Hart
Albrecht W. Hofmann
David E. James
Alan T. Linde
Nancy R. Rice
Richard B. Roberts
Vera C. Rubin
I. Selwyn Sacks
Norbert Thonnard
Kenneth C. Turner
Research Associates
Tom I. Bonner
Mordeckai Magaritz
J. Arthur Snoke
Kiyoshi Suyehiro
Fellows
John R. Bannister
Charles L. Bennett
Arturo Cuyubamba
Gregory S. DeWitt
John R. Evans
Antonio Flores
George H. Pepper
Charles J. Peterson
R. Sundar Rajan
Michael Schneider
Christine M. Seber
Stuart J. Weidenschilling
David J. Whitford
Reports of Departments
and Special Studies
Department of Embryology
Department of Terrestrial Magnetism
Hale Observatories
Department of Plant Biology
Geophysical Laboratory
Developmental Biology Research Group
Department of Embryology
Baltimore, Maryland
James D. Ebert
Director
Donald D. Brown
Director-Designate
Carnegie Institution of Washington Year Book 75, 1975-1976
Contents
Introduction 7
Developmental Genetics by Gene Isolation: The Dual 5S DNA System in Xenopus . . 12
Preparation of 5S DNA recombined with the plasmid PMB9 12
Restriction enzyme maps of two cloned fragments of X. mulleri oocyte-type
5S DNA 14
Characterization of a new 5S DNA from X. mulleri 15
Reconstruction of 5S RNA transcription 16
Sequencing studies on Xenopus 5S DNA 17
The Structure of Ribosomal DNA from Xenopus 20
Studies of Human X-chromosome DNA 22
Chromosomal Proteins and the Control of Gene Function 23
Isolation and characterization of amplified nucleoli 24
The subunit organization of mouse satellite chromatin 27
The Polarity of Ribosomal DNA in Xenopus 28
Determination of polarity with DNA exonucleases 28
Determination of polarity using rDNA transcription complexes 29
Ribosomal DNA in Drosophila melanogaster 30
Purification of Drosophila rDNA 31
Analysis of EcoRI fragments of Drosophila rDNA 32
Length heterogeneity in the spacer region 35
Arrangement of length heterogeneity 35
Distribution of rDNA repeat lengths on the X and Y chromosomes of
D. melanogaster 36
Structural and Functional Studies of Fibroin Genes 36
Physical purification of the gene coding for silk fibroin 37
Cloning of fibroin gene plasmid 37
Accentuated expression of fibroin genes in vivo and in vitro 38
Biogenesis of Mitochondria 45
Cloning of Xenopus laeuis mitochondrial DNA 46
Maps ofX. laeuis andX. mulleri mtDNA: restriction sites, the D-loop and the
rRNA genes 46
Mapping of 4S genes on X. laeuis mtDNA 48
Mapping studies with mtDNA from Drosophila melanogaster 49
Mapping and sequence comparisons of goat and sheep mtDNAs 50
Phospholipid-Cholesterol Dynamics in Mammalian Cells 52
Interactions of phospholipid vesicles with murine lymphocytes: vesicle-cell
adsorption and fusion as alternative pathways of uptake 53
Evidence for a stable vesicle-cell adsorption 53
Vesicle-cell fusion and lipid exchange 56
Interaction of Phospholipid vesicles with murine lymphocytes: vesicle-cell
adsorption leads to altered surface properties and
enhanced proliferative response 57
Enhancement of proliferation by lipid vesicle pretreatment 58
Alterations in lectin-induced agglutinability and lectin-binding
in vesicle-treated cells 59
Alteration of cell surface properties by vesicle adsorption 60
Studies of membrane adhesion 61
Vesicle to cell adhesion 62
Cell to cell adhesion 63
Distribution of lipophilic fluorescent probes in intact cells 66
Membrane Transplantation by Fusion 67
Mechanism of Syngeneic Response Induced by Lectins 73
Studies on Skeletal Muscle and Neuronal Plasma Membranes 76
ACh receptor biosynthesis 78
Autoradiographic analysis of receptor degradation 81
Effect of muscle activity on receptor turnover 83
Organization of muscle plasma membrane in vivo 84
Studies on the mobility of extrajunctional ACh receptors in plasma membrane ... 85
Preparation of a fluorescent derivative of a-bungarotoxin 86
Properties of fluorescent a-bungarotoxin 86
Use of FITC-a-bungarotoxin as a probe in studying membrane structure 87
Turnover of plasma membrane proteins 87
Studies on neuronal plasma membranes 88
Signaling in the Nervous System: Synapses of Specific Neurons in the Leech 90
Chemical synapses in the leech 90
Structural details of the synapse 90
The arrangement of synapses on sensory neurons 91
Sprouting of sensory neuron terminals 91
An electrical synapse in the leech 95
Interruption of the Rohde neuron 96
The Collection of Human Embryos 97
Developmental stages in human embryos 98
Development of the nervous system 98
The Bluntschli Collection 98
Staff Activities 99
Bibliography 102
Personnel 103
INTRODUCTION
In Pentimento* , Lillian Hellman
wrote, "Old paint on canvas, as it ages,
sometimes becomes transparent. When
that happens it is possible, in some pic-
tures, to see the original lines: a tree
will show through a woman's dress, a
child makes way for a dog, a large boat
is no longer on an open sea. That is
called pentimento because the painter
'repented,' changed his mind. Perhaps
it would be well to say that the old con-
ception, replaced by a later choice, is a
way of seeing and then seeing again."
There have been three conceptions of
the Department of Embryology: the
conception of Franklin P. Mall and
George Streeter (1913-1940), the vis-
ion of George W. Corner (1940-1955),
and of the writer (1956-1976). Now
there will be another. By the time this
report is in the hands of its readers, the
Department's fifth Director, Donald D.
Brown, will be charting its course.
The departmental canvas is a large
one, painted and repainted over more
than six decades, not just by four di-
rectors, but by perhaps twenty "mas-
ters" and ten times that number of
their students. Here and there, espe-
cially in the older sections, the paint is
transparent, and one can see and see
again the classic works for which the
Department is justly renowned.
The Department was, for five dec-
ades, the world's leading center for the
study of the human embryo. It
pioneered in the development of pri-
mates for research, having the earliest
successful breeding monkey colony.
Using these animals, large strides
were made toward understanding
menstruation and cyclic changes in the
ovaries and uterus, laying much of the
groundwork for more recent advances
in family planning.
Earlier studies, culminating in Ram-
sey's classic works, set the standard for
^Pentimento: A Book of Portraits, 1973, Little,
Brown & Co., Boston.
research on the primate placenta.
Burns' studies of sex transformation
similarly provided a standard in the
field of sex differentiation. Although
the techniques of cell culture were born
in the neighboring Department of
Anatomy of the Johns Hopkins Medical
School, they came of age in the De-
partment of Embryology especially in
the contributions of M. R. and W. H.
Lewis and, later, Irwin Konigsberg.
During the past two decades the field
of developmental biology has been re-
shaped, as students of development
began to perceive that it should be pos-
sible to understand the factors regulat-
ing the synthesis of specific products
and their assembly during the shaping
of a new organism. The Department of
Embryology has played a part — at
times a key part — in this reshaping
process. However, the paint in these
newer sections of the canvas is not yet
transparent enough for the viewer —
especially one as biased as the
writer — to make out all of the land-
marks. A few do stand out in bold re-
lief. Donald Brown pioneered in the
isolation and characterization of ani-
mal genes and their immediate prod-
ucts, the ribonucleic acids. His study
with Carnegie Fellow John Gurdon of
an anucleolate mutant strain of the
South African clawed toad Xenopus
constitutes a modern classic, offering
decisive proof of the role of the nuc-
leolus in the synthesis of ribosomal
RNAs. By further exploitation of the
anucleolar mutant, Brown was able to
map the genes encoding these RNAs.
He and his co-workers then proved that
during oogenesis the genes for
ribosomal RNAs are amplified into
about 900 copies. His research then
evolved to the point where it became
possible for him to isolate the genes
coding for the 28S and 18S ribosomal
RNA fragments — the first genes to be
isolated from any eukaryotic cell. A
further extension of these techniques
CARNEGIE INSTITUTION
culminated in the isolation and charac-
terization of the genes coding for 5S
ribosomal RNA.
Brown and his co-workers then
turned to the question of whether the
amplification of specific genes can be
demonstrated in specialized cell types.
Together with Suzuki, he isolated the
messenger RNA coding for the silk pro-
tein fibroin in the larval silkworm
Bombyx mori. Using this mRNA as a
probe, Brown was able to show that the
fibroin gene is not amplified in the cells
of the silk glands as compared to other
cells, thereby providing strong evi-
dence that the amplification of specific
genes is not a generalized mechanism
of cell differentiation. Brown's and
Suzuki's studies have provided what
may be the most complete account thus
far available of the waxing and waning
of gene action during development.
The group's continuing investiga-
tions have made a significant impact in
developmental, molecular, and evolu-
tionary biology. Studies of "develop-
mental genetics by gene isolation"
have led to new and brilliantly success-
ful approaches to fundamental ques-
tions in development. The Department
has pioneered in the development and
exploitation of techniques for gene
purification, genes of known function
having been characterized with the
long-term goal of reconstructing their
controls in vitro. Studies of "gene
anatomy," the structure and arrange-
ment of genes, have resulted in new in-
sights into the mechanism by which
multiple tandem genes evolve to-
gether.
It was just a decade ago, in his now-
classic articles in Journal of Molecular
Biology and Proceedings of the Na-
tional Academy of Sciences, that Igor
Dawid proved that the so-called cyto-
plasmic DNA of amphibian oocytes,
long an enigma to embryologists, is in
fact mitochondrial. In cooperation with
a series of able postdoctoral fellows and
students, he has explored in depth the
mitochondrial genes (mtDNA) and
their products, the mitochondrial
ribosomal RNAs (mt-rRNAs) and
transfer (mt-tRNAs).
Cell and tissue interactions are the
hallmark of development. The dif-
ferentiation of most cells requires, at
one or more points in their life his-
tories, interactions with their neigh-
bors, and failures in close range tissue
interactions may result in alterations
in the normal patterns of development.
The question has been posed in a gen-
eral way as follows. How does the mic-
roenvironment make its impact on cell
function via the cell membrane? Genes
direct the synthesis of the components
of cell membranes, and must, in turn,
be regulated by signals from the cell
membrane.
A better understanding of the role of
the microenvironment demanded the
capability to analyze cell differentia-
tion in vitro. Of the Department's
achievements in cellular develop-
mental biology in the 1960s one stands
out: Konigsberg's clonal analysis of
myogenesis. His was the first clonal
study of any differentiating cell. We
are indebted to T. T. Puck of the Uni-
versity of Colorado, who first developed
the "feeder layer" technique, enabling
him and others to clone a number of
established cell lines. It was
Konigsberg, however, who modified
the technique and applied it signifi-
cantly in developmental biology. It de-
serves special mention, too, that two
other members of the Department,
Coon and Kaighn, followed Konigs-
berg's initial success with uncommonly
important clonal analyses of chondro-
cytes and liver, and that a former Car-
negie Fellow, Tokindo Okada, has used
his experience in the Department to
great advantage in his studies of
"transdifferentiation" of pigmented re-
tinal to lens cells.
Another significant advance in cell
biology came in DeLanney and Ebert's
description of the graft-versus-host
reaction. .
But single "major discoveries" do not
tell the whole story. Equally important
is the steady flow of consistent, repro-
ducible findings that, over the years,
illuminate an entire subject. For
nearly two decades the Department
was the Nation's, indeed the world's,
DEPARTMENT OF EMBRYOLOGY
9
leading center of research on the devel-
opment of the heart, with the work of
DeHaan, Rosenquist, and Ebert being
especially noteworthy.
Finally, mention must be made of
the studies of a number of contributors,
notably Kaighn, Lee, Hay, and
Yoshikawa-Fukada, on the interplay of
tumor viruses and differentiating cells.
Their findings provided many of the
"building blocks" in our current under-
standing of viral oncogenesis.
Yoshikawa-Fukada's first description
of the viral oncogenic sequences, lost
sight of at times because of the flood of
related findings in the cancer virus
field, nonetheless marked her as an in-
novating spirit.
1975-1976
In recent years, no single research
program has attracted more attention
to the Department than Douglas Fam-
brough's continuing exploration of the
acetylcholine receptors in skeletal
muscle fiber membranes. Acetyl-
choline (ACh) receptors are of special
interest to Fambrough and his col-
leagues, both as membrane glycopro-
teins and as elements figuring promi-
nently in neuromuscular interactions.
In a series of definitive articles, they
have reported on the appearance of
ACh receptors during muscle develop-
ment, the production and incorporation
of ACh receptors into the surface mem-
branes of myotubes and their sub-
sequent degradation, the clustering of
ACh receptors at newly formed
neuromuscular junctions both in vitro
and in vivo, the number and distribu-
tion of ACh receptors in adult rat
skeletal muscle fibers, and the time
course of change in these parameters
following denervation. They are con-
tinuing to explore physiological aspects
of receptor function, examining further
the biosynthesis and degradation of
ACh receptors and obtaining more ac-
curate estimates of ACh receptor den-
sity on denervated muscle fiber mem-
branes.
In this year's report, Fambrough and
his student Peter Devreotes discuss in
detail their model for the major events
in receptor metabolism. Its main fea-
tures can be summarized briefly. The
protein subunits of which the ACh re-
ceptor is composed are synthesized
with the usual protein synthesizing
mechanisms of the cells and are rapidly
assembled into receptor units resem-
bling those in the plasma membrane.
These newly synthesized receptors
occur upon some membrane systems in
the cytoplasm of the muscle cells and
are collectively referred to as the "pre-
cursor pool." The average residence of a
new receptor in the precursor pool is
2-3 hours. Receptors are transferred
from this pool to the surface by an
energy-requiring process and are then
able to function. The average half-life
of an ACh receptor in the plasma
membrane is 22 hours. Receptors are
degraded by an energy-requiring pro-
cess which involves the internalization
of the receptor and transport to second-
ary lysosomes and then proteolytic de-
struction. Interested readers will find
the evidence supporting these conclu-
sions in the body of the report, along
with indications of the group's new di-
rections, including a foray into the
membranes of neurons.
Richard Pagano and his associates
are also focusing attention on the cell
membrane, their work being concerned
with the possible dynamic role which
membrane phospholipids and choles-
terol may play in the regulation of cell
behavior. During the past several
years they have utilized artificially
generated lipid vesicles to perturb
membrane components in living cells
in a controlled and systematic way. In
the present report they offer findings
on the interactions of phosphatidyl
choline vesicles with murine lympho-
cytes, and the physiological conse-
quences of these interactions. Pagano,
and postdoctoral fellows Huang and
Ozato, have shown that the pathways
of vesicle-cell interactions, enumer-
ated in previous reports (vesicle-cell
fusion, adsorption and lipid exchange)
are exquisitely sensitive to the molecu-
lar composition and physical properties
10
CARNEGIE INSTITUTION
of the artificial vesicles. In the lympho-
cyte system, the resulting membrane
perturbations lead to altered surface
properties and an enhanced prolifera-
tive activity of these cells, when stimu-
lated by various mitogens and anti-
gens.
In a separate line of research, Car-
negie Fellow Masatoshi Takeichi has
explored some of the adhesive prop-
erties of a fibroblastic cell line and re-
ports on his attempts to identify and
characterize membrane proteins which
participate in this phenomenon. Paral-
lel studies on the adherence of lipid
vesicles of known composition to cells
are also discussed.
The importance of cellular interac-
tions is illustrated strikingly in Ken-
neth Muller's studies of synapses in the
leech. The role that a neuron assumes
in the nervous system is dependent
upon the connections that it makes
with other neurons or with peripheral
targets. Therefore, an essential step
both in determining the forces that act
initially to organize the nervous sys-
tem and in characterizing the mecha-
nisms by which a neuron handles, or
integrates, information is to ascertain
the relationship between a neuron's
function and the arrangement of its
synapses. A related problem in
neurobiology has been to determine the
extent that a neuron's shape and other
properties are intrinsic, or whether its
contacts or associations with other
neurons are essential for certain as-
pects of its growth and maintenance.
The nervous system of the leech
Hirudo medicinalis has been particu-
larly favorable for study of the physio-
logical role and functioning of specific
neurons, largely because its organiza-
tion is relatively simple. Neurons in
this segmented worm are grouped 350
together to form essentially identical
ganglia that are linked in a chain by
axon bundles; each ganglion mediates
the reflexes and controls the behavior
of its segment. It has been possible to
monitor electrical activity from within
single sensory and motor neurons with
microelectrodes and to correlate the
physiological activity of functionally
identified neurons with their charac-
teristic shape and position within each
ganglion. By recording from pairs of
neurons one can trace synaptic connec-
tions, which have been found to be sur-
prisingly uniform and in that sense
predictable between specific neurons.
In his current report, Muller de-
scribes the structural details of chemi-
cal synapses, delineates their ar-
rangement on sensory neurons, and
explores the sprouting of sensory
neuron terminals. He also describes
the unexpected discovery of an electri-
cal junction between unpaired Rohde
neurons in neighboring ganglia.
Brown and his colleagues are now
concentrating on the family of genes
which code for the small RNA molecule
of ribosomes termed 5S RNA. The
"transcription unit," that is the region
of this DNA which is transcribed into
RNA in the living cell, is relatively
simple — only 120 base pairs in length.
Moreover, the form of RNA polymerase
which transcribes this gene in vivo has
been identified. The fact that the ma-
ture RNA transcript does not appear to
be derived from a higher molecular
weight precursor is especially impor-
tant: It means that the true initiation
site of transcription from 5S DNA is
known. Therefore, one can detect faith-
ful transcription when it occurs in
vitro. Using this system Brown hopes
to reassemble the macromolecules
which enable this DNA to be trans-
cribed with fidelity in the living cell. In
this way he hopes to elucidate the con-
trol mechanisms of an animal gene.
This year his group reports progress
along several lines. Three different
kinds of 5S DNA have been cloned \r\E.
coli by means of the new recombinant
DNA methodology, thereby providing
an order of magnitude more material to
work with and permitting the careful
examination of one single repeating
unit of a tandem gene family. The "re-
DEPARTMENT OF EMBRYOLOGY
11
striction maps" of these cloned DNAs,
i.e., the sites where they are cleaved by
certain restriction enzymes, are de-
scribed by Fellows Jeffrey Doering and
Scott Emmons. The latter is attempt-
ing to develop an assay for proper
transcription. Fellow Nina Fedoroff is
exploring DNA sequencing methods to
determine the exact nucleotide se-
quence in the spacer regions of 5S
DNA, particularly those regions adja-
cent to the initiation site of the gene.
DNA structure continues to hold the
attention of several Staff Members and
Fellows. Reeder and Dawid, along with
Fellow Peter Botchan, report on the
structure of nontranscribed spacer re-
gions in Xenopus ribosomal DNA,
based on studies using cloned frag-
ments of rDNA generated with the re-
striction endonuclease EcoRI. Fellow
Peter Wellauer and Dawid have under-
taken the first stages of a study of the
ribosomal DNA of Drosophila, Dawid
and his colleagues are cloning Xenopus
mitochondrial DNA, and Carnegie Fel-
low Ohshima and Staff Member Suzuki
are attempting to clone the silk fibroin
gene.
In all of these studies, the Depart-
ment has adhered to the safety stan-
dards proposed by the Asilomar Con-
ference on Recombinant DNA.
Mention has already been made of
the contributions of several fellows
and students. Peter Wellauer, whose
time in 1975-1976 was shared between
California Institute of Technology and
this Department, will return to Swit-
zerland to a new position in the Swiss
Cancer Research Institute in
Lausanne. Scott Emmons will continue
his postdoctoral training at the Uni-
versity of Colorado. Leaf Huang has
taken up his new responsibilities in the
Department of Biochemistry of the
University of Tennessee, and
Katherine Tepperman will join the
University of Cincinnati's faculty.
Keiko Ozato will move to the Johns
Hopkins School of Medicine-Good
Samaritan Hospital.
Continuing in the Department are
Diana Card, a Fellow of the Muscular
Dystrophy Associations of America,
Inc., with Fambrough; Peter Botchan
and Harvey Wahn, with Reeder;
USPHS Fellows Jeffrey Doering and
Nina Fedoroff with Brown; Paul
Geshelin and Yasumi Ohshima, with
Suzuki; Alex Sandra, with Pagano;
Elizabeth Godwin, with Dawid; and S.
T. Carbonetto, with Muller and Fam-
brough.
This brief listing of postdoctoral fel-
lows beings me full circle to my start-
ing point in 1956. In his autobiog-
raphy, Anatomist at Large, George
Corner wrote that the secret of his suc-
cess was his ability to place himself
under wise leaders. His words ring true
for me as well. I would emphasize still
another ingredient: the catalysis pro-
vided by young minds — fellows and
students. When President Haskins and
I began to chart the Department's
course in 1956, a vital principle
emerged. It was clear that even if
ample funds and space were available
immediately to undertake all the tasks
we had set for ourselves, the scientific
talent was not. The pool of able devel-
opmental biologists then was small. It
was agreed, therefore, that training
would henceforth be an integral part of
the Department's program. The wis-
dom of that decision is demonstrated by
a glance at today's staff — Brown,
Dawid, Fambrough, Reeder and Suzuki
all joined the Department initially for
training. Thus the reshaping of the
Department has been gradual, not
cataclysmic. As I wrote in 1956 (Year
Book 55, p. 270), the Department may
be likened to an organism. Although
there has been considerable change or
turnover within its parts, as a whole it
enjoys stability. We achieve what
Rudolph Schoenheimer called a condi-
tion of stability with flux — working
toward a better understanding of the
mechanisms of development, using for
that purpose whatever methods be-
come necessary.
12
CARNEGIE INSTITUTION
DEVELOPMENTAL GENETICS BY GENE ISOLATION: THE DUAL
5S DNA SYSTEM IN Xenopus
D. D. Brown, J. Doering, S. Emmons, N. Fedoroff, and E. Jordan
In Year Book 74 (p. 12) we described
our reasons for concentrating on the
family of genes which code for the
small RNA molecule of ribosomes
termed 5S RNA. The "transcription
unit," that is, the region of this DNA
which is transcribed into RNA in the
living cell, is relatively simple — only
120 base pairs in length. The form of
RNA polymerase which transcribes
this gene in vivo has been identified by
Roeder as form III. The mature RNA
transcript does not appear to be derived
from a higher molecular weight pre-
cursor. This observation is especially
important because it means that we
know the true initiation site of trans-
cription from 5S DNA and therefore
can detect faithful transcription when
it occurs in vitro. Using this simple
system we hope to reassemble the mac-
romolecules which enable this DNA to
be transcribed with fidelity in the liv-
ing cell. In this way we hope to eluci-
date the control mechanisms of an
animal gene.
In this report we summarize progress
made along several lines. Three dif-
ferent kinds of 5S DNA have been
cloned in E. coli by means of the new
recombinant DNA methodology (see
Year Book 74). This approach provides
an order of magnitude more material to
work with and permits us to examine
carefully a single repeating unit of a
tandem gene family. We describe the
"restriction maps " of these cloned
DNAs, i.e., the sites where they are
cleaved by certain restriction enzymes
(Doering and Emmons). Various
schemes are being investigated to de-
velop an assay for proper transcription
(Emmons). DNA sequencing methods
will determine the exact nucleotide se-
quence in the spacer regions of 5S
DNA, particularly those regions adja-
cent to the initiation site of the gene
(Fedoroff). We have two new species of
Xenopus obtained from the collection of
M. Fischberg of Geneva, Switzerland.
We have purified partially the princi-
pal 5S DNA component from each of
these. Although we will not describe
these experiments in this report, we
have a specific purpose for isolating the
genes from different species which en-
code the same function. The spacer re-
gions that separate the genes have di-
verged considerably in their nucleotide
sequences. Control regions within
these spacers are likely to be con-
served. By comparing the nucleotide
sequences in comparable regions of 5S
DNA isolated from different species we
can identify possible control regions.
Preparation of 5S DNA
Recombined with the Plasmid
PMB9
D. D. Brown and E. Jordan
We have used exclusively the plas-
mid PMB9 for recombining with, and
cloning, Xenopus 5S DNA. This circu-
lar plasmid has a molecular weight of
3.5 x 10 6 (5400 base pairs). It has a
single site for several restriction en-
zymes into which foreign DNA can be
introduced and joined. We have used
the plasmid site that is cleaved by the
enzyme Hindlll for insertion because
three different Xenopus 5S DNAs also
have Hindlll sites. This enzyme leaves
at each end four nucleotide single-
stranded regions that are complemen-
tary and can be joined by ligase.
Xenopus laevis oocyte-type 5S DNA
has a single Hindlll site once per re-
peat, and Xenopus mulleri oocyte 5S
DNA has irregular Hindlll sites which
occur on the average about once every
four repeats. The newly isolated 5S
DNA from Xenopus mulleri also has a
single Hindlll site per repeat. The
scheme for insertion is as follows. The
plasmid is digested to completion with
Hindlll. The 5S DNA is digested either
DEPARTMENT OF EMBRYOLOGY
13
completely or partially with Hindlll,
depending upon whether monomers or
multimers are to be inserted. The
plasmid and 5S DNAs are mixed and
joined with T4 ligase. The bacteria are
transformed with the recombined
DNA. Those bacteria which incorpo-
rate a plasmid can be selected by their
resistance to colicine or tetracycline.
Grunstein and Hogness (Proc. Nat.
Acad. Sci. 72, 3961, 1975) devised a
simple screening test to detect bacteria
that contain a specific inserted gene.
Many individual colonies are grown on
a filter paper disc and their DNA is
immobilized on the paper after al-
kaline denaturation. The filter is then
hybridized with iodine- 125 labeled 5S
RNA. Only the clones that contain 5S
DNA hybridize with the 5S RNA and
become radioactive. The radioactive
clones are located by exposing the fil-
ters to x-ray film. Individual clones are
grown up for analysis. Figure 1 demon-
strates one such screening experiment.
This method of colony hybridization
not only simplifies the identification of
bacteria that contain 5S DNA sequen-
ces, but it avoids growing large amounts
of bacteria that might contain un-
known DNA sequences. Such bacteria
remain on the agar plate and are auto-
claved. Since 5S DNA occurs as a tan-
dem gene family, no other DNA
molecules are likely to be inserted into
a plasmid with the 5S DNA fragment.
These considerations are important be-
cause of our continuing concern with
safety precautions, especially when
there is a possibility of cloning un-
known animal DNA fragments. The
colony isolation method described
above minimizes this possibility. In
addition we have introduced a "chemi-
cal" containment step during the isola-
tion of recombinant DNA. When we
grow E. coli cultures containing a re-
combined plasmid, we kill the bacteria
before isolating their plasmid DNA.
This step is achieved simply by adding
small amounts of chloroform to the cul-
tures just before harvesting. No viable
bacteria remain after five minutes of
chloroform treatment.
Fig. 1. Autoradiograph of bacterial colonies
that had been hybridized with 1J3 I-labeled 5S
RNA. Tetracycline-resistant colonies (containing
the plasmid or plasmid plus 5S DNA insert) were
grown on the filter paper, treated with alkali to
denature DNA, neutralized, dried, and hy-
bridized with the radioactive RNA. The excess
radioactivity was removed, and the labeled col-
onies containing 5S DNA molecules were de-
tected by autoradiography. Top, a photograph of
three filter papers with dried colonies. The mid-
dle filter contains bacteria that had been trans-
formed with PMB9 recombined with X. laevis
oocyte-type 5S DNA. Bacteria on the other two
filters were transformed with PMB9 recombined
with X. mulleri oocyte-type 5S DNA. The film
itself is shown in the middle panel, and it has
been placed over the filters to show which ones
contain the 5S DNA insert (bottom panel).
14
CARNEGIE INSTITUTION
The Structure of X. laevis Oocyte-Type
5S DNA
A model for one repeating unit of X.
laevis oocyte-type 5S DNA (XLo) was
first published in Year Book 74. This
model is refined and corrected in Fig. 2.
The A+T-rich spacer region was placed
erroneously at the 3' end of the trans-
cription unit (the 5S RNA gene). We
know that there is a short A + T-rich
region adjacent to the 3' end of the
gene from the experiments of R. D.
Brown {Year Book 74). However, the
recent sequencing studies by Claude
Jacq and George Brownlee at the Med-
ical Research Council Laboratory of
Molecular Biology in Cambridge, Eng-
land, have shown that the principal
A+T-rich spacer region is at the 5' end
of the transcription unit. Our current
model (Fig. 2) incorporates this change
in orientation. The A+T-rich 15-mer
oligonucleotides that comprise most of
this region are designated by short dot-
ted lines. Their exact location and
number are unknown. Jacq and
Brownlee also partly sequenced an in-
teresting region called the "pseudo-
gene," which differs from the gene se-
quence by only six bases in the region
from residue 45 to residue 100. The
terminal 20 nucleotides are completely
different; sequencing of the first 44
nucleotides is incomplete. The recogni-
tion site for the restriction enzyme
Haelll (Fig. 2) occurs in a part of the
pseudogene which is analogous to a
corresponding region in the gene. The
other Haelll sequence found in the
gene is missing in the pseudogene. The
function of the pseudogene is unknown.
Restriction Enzyme Maps of Two
Cloned Fragments ofI. mulleri
Oocyte-Type 5S DNA
J. Doering and S. Emmons
The restriction enzyme Hindlll cuts
X. mulleri oocyte-type 5S DNA (Xm )
occasionally yielding fragments which
are heterogeneous in length. Frag-
ments from a complete Hindlll digest
of Xm were cloned by the methods de-
scribed above in E. coli HB101 using
the plasmid PNB9. Two of these cloned
fragments were selected for detailed
study.
Xm 3 is a 2700 base-pair fragment.
It is cut by the restriction enzyme
Hpall into four fragments, the smallest
of which is present in two copies. To
determine the order of these frag-
ments, Xm 3 was end-labeled using
a- 32 P-labeled ATP and T4 polynuc-
leotide kinase. The labeled DNA was
digested with the enzyme and the
fragments run on a gel. Radioactive
terminal fragments were identified by
autoradiography. The order of the re-
maining fragments was deduced from
the sizes of partial digestion products
(Fig. 3). The pattern of fragments ob-
tained after double digestion with
Hpall and Haelll was indistinguish-
able from that obtained with Haelll
alone, indicating that all Hpall sites
are very close to Haelll sites. This im-
plies that Hpall cleaves only within
the gene sequence itself, since the rec-
ognition sequence for Hpall occurs
next to an Hae site in X. mulleri
oocyte-type 5S RNA. The restriction
map in Fig. 3 was constructed using
A + T regi on
Hind
nr
G + C region
Hind
Gene Pseudogene HE
*-$'
327- 507-
Hae Hae
->+« 120
Hae
-103 -H
Fig. 2. Model of one repeat ofX. laevis oocyte-type 5S DNA. Restriction enzyme sites are included.
The numbers refer to base pairs.
DEPARTMENT OF EMBRYOLOGY
15
Hind
m
Hind
m
->!&*-
+*fi 190*^190^7*-
190 100
t=Hpa H
I =Hae m
: EHae HI (UNCERTAIN)
Fig. 3. Restriction map of oneX. mulleri oocyte-type cloned fragment (Xm„3). The numbers refer to
base pairs. Some cleavage sites are tentative. They are shown as dotted lines.
end-labeled fragments, partial digests,
and Haelll digests of each Hpall frag-
ment.
Xm l is an 800 base-pair fragment
containing one Hpall site, one Hhal
site, and six Haelll sites. A map is
being constructed using the methods
described above.
Characterization of a New 5S
DNA fromI. mulleri
J. Doering
In the search for the 5S DNA coding
for the somatic-type 5S RNA, new 5S
DNA-containing satellites were iso-
lated from both X. laeuis and X. mul-
leri. The new X. laevis 5S DNA was
found to code for a 5S RNA different
from both the oocyte and somatic types
(Year Book 74, p. 13).
The kind of 5S RNA gene carried by
the newX mulleri 5S DNA is still un-
known. Table 1 lists some characteris-
tics of Xm-new DNA compared with
those of the oocyte-type 5S DNA from
X. mulleri. The partially purified new
5S DNA from X. mulleri (Xm-new) is
cut by the restriction enzyme Hindlll
into homogeneous 900-base-pair frag-
ments (Fig. 4). Higher molecular
weight DNA is also present after diges-
tion, but only the 900-base-pair frag-
ments hybridize with 5S RNA. Frag-
ments from a complete Hindlll digest
of Xm-new were cloned by the methods
described above. The Xm-new DNA
from one clone has been subjected to
detailed study.
Fig. 4. Electrophoresis in a 2% acrylamide,
0.5% agarose composite gel of Hindlll digests of
Xm-new 5S DNA (left), purified from the ani-
mal's DNA, PMB9 containing one Xm-new 5S
DNA insert (middle), and a mixture of these two
samples (right). Electrophoresis is from top to
bottom.
16
CARNEGIE INSTITUTION
TABLE 1. Characteristics of the Newly Isolated 5S DNA fromX. mulleri
Compared toX. mulleri Oocyte 5S DNA
Oocyte-5S DNA
New 5S DNA
Density (g/cm :i ) 1.695
1.712
Average no. of repeats 9000
700
Repeat length (base pairs) 1400
900
Length heterogeneity great
not detectable
Strand separation in alkaline CsCl yes
yes
5S RNA coding strand light
light
Restriction Enzyme Sensitivity (cuts/repeat)
EcoRI resistant
resistant
Hindlll less than one
1
Hpall variable
3
Hhal variable
2
Haelll variable
> 5
The cloned Xm-new DNA has a
biphasic melting curve, and both re-
gions are high in G+C content. Haelll,
which cuts DNA at the sequence 5'-
GGCC-3', digests about one-third of
the 900-base-pair fragments into pieces
of less than 40 base pairs. This
suggests that the spacer has a short re-
peating sequence that contains a re-
petitive Haelll cleavage site. The
spacer region of cloned Xm-new DNA
has little, if any, homology with Xm 3
DNA by hybridization criteria. While
E. coli RNA polymerase was found to
transcribe the noncoding strand of X.
laeuis oocyte-type 5S DNA preferen-
tially, the enzyme transcribes both
strands of cloned Xm-new DNA about
equally.
A restriction enzyme map of the
cloned Xm-new DNA is being con-
structed in preparation for sequencing
studies to determine what type of 5S
gene the new DNA carries.
Reconstruction of 5S RNA
Transcription
S. Emmons
Our efforts to devise a system in
which to detect and study 5S RNA
transcription in vitro {Year Book 74, p.
19) have continued over the past year.
The immediate aim is to develop a sen-
sitive and specific assay for the initia-
tion of transcription at the beginning of
the 5S RNA gene.
For this purpose we have available to
us a number of small (one hundred to a
few hundred base pairs), well-
characterized DNA "restriction frag-
ments" cut from various regions of 5S
DNA by restriction enzymes. Some of
these fragments carry the initiation
site for 5S RNA transcription, being
produced by a cut before the gene re-
gion and a cut within the gene region.
We envisage that these fragments may
be used in one of two ways in transcrip-
tional studies. First, they may be used
as hybridization probes to determine
whether a population of RNA trans-
criptions contains any molecules whose
5' ends are identical to the 5' end of 5S
RNA. In this method, RNA transcripts
are hybridized to the restriction frag-
ment and treated with a single-
strand-specific ribonuclease. RNA
molecules resulting from a correct in-
itiation event will give rise to an RNA
fragment equal in length to the known
length of the gene region carried on the
restriction fragment. This RNA frag-
ment could be identified by gel elec-
trophoresis and nucleotide sequencing.
In the second method, the DNA restric-
tion fragment is used as a template for
RNA polymerase or for RNA-
synthesizing extracts. Since the frag-
ment ends within the gene, correct in-
itiation of transcription again results
in an RNA fragment of known length.
Both of these approaches have been
attempted using the (approximately)
450 base-pair fragment produced by
DEPARTMENT OF EMBRYOLOGY
17
the restriction enzyme Haelll from X.
laeuis oocyte 5S DNA. This fragment
carries all of the A + T-rich spacer, the
hypothetical promotor region(s) before
the gene, and the first eight nuc-
leotides of the gene itself (Fig. 2). To
determine whether this fragment could
be used as a hybridization probe, at-
tempts were made to hybridize it with
125 I-labeled 5S RNA. For this purpose,
the fragment was denatured and bound
to nitrocellulose filters, and RNA an-
nealed to it under conditions favoring
the stability of the very short hybrid ( 1
M NaCl, 0°C). Under these conditions a
13-nucleotide-long oligomer (produced
from 5S RNA by ribonuclease Tl) an-
nealed to 5S DNA, but shorter
oligonucleotides did not, and 125 I-5S
RNA did not anneal to the restriction
fragment. It is presumed that this is
because the RNA was not at a high
enough concentration. Unfortunately,
higher concentrations are not obtaina-
ble easily in an assay using a mixture
of transcripts, so it is concluded that if
restriction fragments are to be used as
hybridization probes they must carry
more than the first eight nucleotides of
the gene.
The 450-nucleotide Haelll fragment
from X. laevis has also been used as a
template in transcription studies. A
protein extract of X. laevis ovaries was
prepared following procedures of
Roeder (J. Biol. Chem. 249, 241-248,
1974; personal communication), and its
RNA polymerase activity analyzed
using various templates. This extract
was found to be relatively inactive on
Xenopus 5S DNA fragments, compared
with poly dAT and withE". coli plasmid
fragments. No activity at all (conver-
sion of precursor to TCA insoluble ma-
terial) could be observed even with
high concentrations of the 450 nuc-
leotide fragment. In one experiment,
the reaction mixture was nevertheless
fractionated in a system (homo-
chromatography on DEAE paper) that
would have revealed the presence of an
eight nucleotide oligomer, but none
was seen.
The 5S gene system of Xenopus re-
mains an optimal one for attempts to
reconstruct eukaryotic transcription in
vitro. Purified 5S DNAs from several
species are available as templates, and
the structure of these DNAs is being
studied intensely. Two classes of genes,
expressed in different tissues (somatic
and oocyte) exist, allowing an analysis
of the mechanisms operating to control
their expression. In addition, the pri-
mary transcription product in this sys-
tem is small and of known sequence.
Sequencing Studies on
Xenopus 5S DNA
N. Fedoroff
We have focused on the introduction
and development of direct DNA
sequencing techniques. DNA sequenc-
ing methodology has been evolving
rapidly over the past few years. The
approach that has proved most fruitful
is identification of the terminal nuc-
leotide in successively shorter frag-
ments having one fixed end. With
technical improvements in oligonuc-
leotide fractionation procedures, the
number of nucleotides that can be or-
dered in a single analysis has increased
steadily. Several laboratories are en-
gaged in developing and improving dif-
ferent methods for generating the req-
uisite fixed-end fragments and iden-
tifying terminal nucleotides.
Our immediate interest is in the nuc-
leotide sequence of the spacer region
adjacent to the 5 ' terminal portion of
the 5S RNA gene in Xenopus, since this
is likely to contain a control region for
transcription. All of our studies so far
have been done on a single repeating
unit of X. laevis oocyte-type 5S DNA.
Quantities sufficient for sequencing
studies have been obtained by integra-
tion of the DNA into bacterial plasmids
as described in previous sections. The
distribution of restriction endonuclease
Haelll sites within this DNA, shown in
Fig. 2, makes the region of interest ac-
cessible to direct sequencing in several
ways.
We have, for example, used two dif-
18
CARNEGIE INSTITUTION
t, if
T 3 5 T
5' A + T - r ic h region 3
5S gene
DN A poly mera se
a-P 32 d NTP's
HaeUr site
WAAAA/VWAAV**-
Hae n digest ion
denat lira t ion
Hoe HI site
V r 3
A+T- rich region 55 gene
X Pol y nucle
ir DNase I
7
tide kinase
P" AT P
exonucl ease
,Hae HI site
denaturation
\A/VWWWVA/<AA«.
\aava\*w
Fig. 5. Two schemes for direct DNA sequencing applied to the region adjacent to the initiation site
of 5S DNA. A series of labeled DNA fragments (wavy lines) are generated which have a unique 5'
terminus and a known 3'-end; that is, the nucleotide at the 3'-end is known to be A, T, G, or C. These
different-sized fragments are separated by polyacrylamide gel electrophoresis, (a) The Sanger-
Coulson method, (b) A modification developed in this laboratory. Details are given in the text.
ferent procedures to generate frag-
ments which have a common 5 ' ter-
minus and which extend over the re-
gion adjacent to the gene (Fig. 5). Fig-
ure 5a represents a direct application
to 5S DNA of the scheme developed by
Sanger and Coulson (J. Mol. Biol. 94,
441_448, 1975). The strand of 5S DNA
containing the 5S RNA sequence (non-
coding strand) is used as a template for
E. coli DNA polymerase I. The Haelll
fragment extending from residue 9 to
residue 66 of the 5S RNA sequence
(Fig. 2) is used as a primer for the en-
zyme. The strand of the fragment com-
plementary to 5S DNA (coding strand)
is annealed to the template extended
with DNA polymerase in the presence
of cv- :i2 P-labeled nucleotides. This gen-
erates a population of labeled frag-
ments that begins at residue 8 of the 5S
gene and extends into the adjacent
spacer region (Fig. 5a). The identity of
the terminal nucleotide in the labeled
fragments is determined by the follow-
ing procedure. The extended primer-
template complex is purified and
aliquots are reincubated with DNA
polymerase in the presence of either
one or three unlabeled nucleotides.
Under the conditions used, the labeled
fragments are degraded until they
terminate at the single included nuc-
leotide, or extended until they termi-
nate just prior to the single omitted
nucleotide. For example, the reaction
mixture containing TTP will yield only
those oligonucleotides terminating in T
residues. Since all fragments have a
common 5 ' end, the position of T resi-
dues in the sequence can be deduced
directly from the length of the
T-terminated fragments. Separation of
oligonucleotides that differ in length
by a single residue is carried out by
electrophoresis in high-percentage de-
DEPARTMENT OF EMBRYOLOGY
19
naturing poly aery 1 amide gels.
One difficulty with this approach is
inherent in the organization of the 5S
DNA. As described in an earlier sec-
tion, each repeating unit of Xenopus
laeuis oocyte 5S DNA contains both a
gene and a very similar sequence,
termed the "pseudogene" (see Fig. 2).
The gene and pseudogene differ by ap-
proximately four nucleotides within
the segment covered by the primer
molecule. The primer, therefore, hy-
bridizes to both regions, and the ex-
tended primer molecules represent se-
quences adjacent to both gene and
pseudogene. One way around this diffi-
culty is to determine the sequence of
the short fragment released by digest-
ing the extended template-primer
complex with Haelll as shown in Fig.
5a. Since the pseudogene lacks this
Haelll site, the radioactive fragments
will be released from primer molecules
on gene sequences but not pseudogene
sequences. In principle, this permits
determination of at least the first 50
residues of the spacer sequence adja-
cent to the gene before the longer
fragments extended from the
pseudogene interfere. Using this ap-
proach, we have obtained useful se-
quence information over a 30-
nucleotide region beginning between
10 and 20 nucleotides from the first in-
tragenic Haelll site. The first 10 to 20
nucleotides are recovered in very low
yield or not at all because of differen-
tial labeling and probably also because
of differential degradation of short
oligonucleotides.
We have developed an alternative
procedure for generating the same
population of labeled fragments that
includes the region which is adjacent to
the gene. In this procedure, illustrated
in Fig. 5b, the population of labeled
oligonucleotides is derived directly
from the Haelll fragment of 5S DNA
which contains the A+T-rich spacer re-
gion and extends to the 8th residue of
the gene sequence (Fig. 2). The coding
strand of the fragment is labeled at the
5' end using a- 32 P-labeled-ATP and
polynucleotide kinase. Label is intro-
duced in only one of the two strands,
essentially as described by Maniatis et
al. (Proc. Nat. Acad. Sci., U.S.A. 72,
1184, 1975). A larger Haelll fragment
containing a Hindlll site is labeled at
both ends and cleaved with Hindlll.
The two resulting fragments, each
labeled in only one strand, are recov-
ered separately. The labeled fragment
of interest is then nicked internally
with DNase I at a low level. The nicks
are enlarged into gaps by the 3 ' to 5 '
exonuclease activity of either the T4
DNA polymerase or snake venom
phosphodiesterase. The gapped com-
plex is equivalent to the extended
primer-template complex in Fig. 5a,
and subsequent steps in the sequencing
procedure are identical. The fragments
are extended with DNA polymerase in
the presence of either one or three of
the cold nucleotides and the lengths of
the specifically terminated fragments
are determined on a gel. As with the
previous method, we observed a dif-
ferential loss of short oligonucleotides.
Using a fragment of known sequence
(the intragenic Haelll fragment), we
have been able to derive an accurate
sequence to within about 20 nuc-
leotides from the 5 ' end of the
molecule. Another difficulty we have
encountered is that certain nicks in-
troduced by DNase I appear to be rela-
tively resistant to one or the other of
the exonucleases used.
In general, the strengths and weak-
nesses of these enzymatic approaches
are those discussed by Sanger and
Coulson. The principal advantage of
this method is that each terminal nuc-
leotide is identified separately and
each sequence is, in effect, determined
twice within the same experiment. A
shortcoming of this method is in
sequencing runs of the same nuc-
leotide, since a run is usually repre-
sented only by the oligonucleotide ter-
minating just before the run and the
oligonucleotide containing all mem-
bers of the run.
In addition to the enzymatic methods
20
CARNEGIE INSTITUTION
described above, we are using a chemi-
cal sequencing method for DNA which
has been developed in the laboratory of
W. Gilbert. In this procedure, termi-
nally labeled DNA fragments are
cleaved chemically at specific bases,
followed by the gel electrophoretic sep-
aration described above.
THE STRUCTURE OF RIBOSOMAL DNA FROM Xenopus
R. H. Reeder, P. Botchan, and I. B. Dawid
The nontranscribed spacer regions in
Xenopus rDNA are heterogeneous in
length; the physical basis for this
length heterogeneity was described in
Year Book 74 (pp. 19-36). The studies
of spacer structure involved the use of
cloned fragments of rDNA generated
with the restriction endonuclease
EcoRI. EcoRI cuts each repeating unit
of rDNA (gene plus spacer) into just
two fragments — one containing only
gene sequences and a second fragment
that contains all of the nontranscribed
spacer region plus some gene region on
either end. Four of these spacer con-
taining fragments (each with a dif-
ferent length spacer) have been cloned
using a bacterial plasmid as vector.
These cloned EcoRI fragments were
used for the heteroduplex mapping ex-
periments reported last year and have
also been used in the restriction map-
ping described here. The structure of
one of the fragments, Xlrl4, is shown
in Fig. 6.
28S a B
500 f 900 800
430
760 660 660 660
BAM
Fig. 6. Structure of the cloned rDNA fragment
XIrl4. This fragment was generated by cleavage
with EcoRI. As shown on the upper map, it con-
tains some transcribed gene region on each end
and all of the nontranscribed spacer region in the
middle. The location of regions A, B, C, and D
were defined from heteroduplex mapping. The
dotted lines indicate regions that contain repeat-
ing sequences. The lower map shows the location
of BAMHI cutting sites on the same fragment.
Molecular weights are in base pairs.
Measurement of heteroduplexes
formed between long and short cloned
spacers led to the conclusion that each
spacer contains two regions, each com-
posed of a short repeating nucleotide
sequence. One such region (region B in
Fig. 6) is near the 28S sequence and the
other (region D) is near the 18S se-
quence. Regions B and D appear to be
separated from each other by a non-
repetitious sequence (region C). In
order to gain more insight into the
structure of the nontranscribed spacer,
we have begun to dissect these cloned
fragments with other restriction en-
zymes.
The procedure we have used for
mapping restriction cuts is as follows.
Cloned spacer-containing fragments
were labeled at both ends of the
molecule using polynucleotide kinase
to put a 52 P atom on the 5' end of each
DNA chain. The label at the 28S end
can be eliminated by restriction with
endonuclease Hindlll, which cleaves
once, 500 bases from the 28S end. If the
fragment now labeled only at the 18S
end is partially cleaved with a restric-
tion endonuclease, an overlapping set
of fragments is created, all possessing
the same :i2 P labeled 18S termini. If
the partial digestion products are then
electrophoresed on a 1% agarose gel
and their positions determined by au-
toradiography, the order of the termi-
nal digestion products can be deter-
mined by subtracting the molecular
weight of one partial product from the
molecular weight of the next largest
partial product. This procedure has
been used for ordering the fragments
produced by endonucleases Smal, from
Serratia marcescens, and BAMHI, from
Bacillus amyloliquefaciens. Results of
DEPARTMENT OF EMBRYOLOGY
12 13 14 15
6450-
5150-
4000 -
2420-
1430-
"ip(w-
860
500- W Jf «* m -
Fig. 7. Autoradiographs of partial Smal digests
of four cloned spacer containing fragments. Four
EcoRI-generated spacer fragments (Xlrl2, 13,
14, and 15) were end-labeled with polynucleotide
kinase, digested with Hindlll to remove the 28S
end, and then partially digested with Smal. After
electrophoresis on 1% agarose the gels were
dried and autoradiographed. The bands at 5150
and 6450 are from ColEl plasmid carrier.
Molecular weights are in base pairs.
Smal partial digests on four cloned
spacer-containing fragments are
shown in Fig. 7 (the 500-base-pair ter-
minally labeled fragment has not been
separated out and appears as the
bottom-most band). The Smal diges-
21
tions reveal that the fragments pro-
duced from the 18S end of all clones are
identical up to a point 2420 base pairs
from the 18S end. At this point the lo-
cation of Smal sites diverges sharply
among the different clones. The loca-
tion of the divergence boundary agrees
well with previous estimates of the
gene-spacer boundary made from
RNA-DNA heteroduplex measure-
ment. Therefore, it appears that the
nontranscribed spacer region shows
some divergence close up to the gene-
spacer interface. Within the transcrip-
tion region, however, no sequence var-
iation has been found among the four
clones.
Smal also appears to recognize a re-
peating sequence in region B. The
clearest example is Xlrl2 in Fig. 7
where Smal produces a series of finely
spaced bands starting at 4000 bases
from the 18S end.
Mapping of the restriction fragments
produced by BAMHI reveals that this
enzyme recognizes sites only within
the spacer region of all four clones.
With three of the four clones BAMHI
produces a fragment of about 3300 base
pairs containing the 18S end, and from
the same clones a fragment of about
2100 base pairs containing the 18S
end. Internal to these terminal frag-
ments is a region of expanding size
which accounts for most of the size
heterogeneity of the four clones. The
internal spacer fragments vary from
about 500 base pairs on the smallest
clone to a 660 base-pair fragment that
tandemly repeats itself five times on
the largest clone. The exact order of
BAMHI fragments on cloned Xlrl4 is
shown in Fig. 6.
Further restriction analysis has been
done on the 2100 base-pair BAMHI
fragment containing the 28S end. This
fragment overlaps region B (Fig. 6)
which has been shown by electron mic-
roscopy to contain a highly repeating
sequence 100 base pairs or less in size.
Digestion of the BAMHI fragment with
either of the restriction endonucleases
22
CARNEGIE INSTITUTION
Hpall, RII, Smal, HHA or Haelll re-
veals that all recognize a higher order
repeating sequence of about 250-300
base pairs in length. Since all these en-
zymes recognize either the sequence 5 '
CCCGGG 3 ' or a simple variation of
this, it is possible that the basic small
repeat in this region is closely related
to this sequence. In summary, the re-
striction analysis when combined with
the electron microscopy reported last
year indicates that: (a) the length
heterogeneity of the clones is largely
confined to one region of the spacer,
near the 28S end (region D; Fig. 6); (b)
the spacer is composed of a short re-
peating sequence that is probably 100
base pairs or less in size; (c) the D re-
gion and possibly the B region also ex-
hibit a higher order repeating struc-
ture superimposed upon the smaller
sequence repeat. The latter two obser-
vations are characteristic of all "satel-
lite" DNAs so far analyzed with re-
striction enzymes. The interest in this
case is that the satellite is regularly
interspersed between repeated gene
sequences.
STUDIES OF HUMAN X-CHROMOSOME DNA
B. Migeon
We have begun attempts to isolate
the human X-chromosome or portions
of it. The X-chromosome in mammals
is subject to a control mechanism
termed heterochromatinization in
which all but one X-chromosome in a
cell are in a condensed, inactive form in
interphase cells. Although there has
been much speculation, little is known
about the molecular basis for this
unique form of gene regulation.
In metaphase, inactive X-chromo-
somes (although indistinguishable
from their active homologues by all
chromosome staining techniques) can
be identified because they replicate
very late during DNA synthesis and
are the last chromosomes to terminate
replication. The inactive X is also iden-
tifiable in interphase nuclei because of
- 1.750
-1.700
650
600
10 15
Fraction Number
Fig. 8. BrdU-labeled DNA from isolated chromosomes centrifuged to equilibrium in CsCl. The
purified chromosome fraction of human lymphocytes that had been stimulated to divide with a
mitogen was labeled with :1 H-BrdU for 1 hr and chased for 3 hr. cts/min (x x) A. (1( , ( ).
DEPARTMENT OF EMBRYOLOGY
23
its highly condensed state at a time
when the majority of DNA from other
chromosomes is dispersed. It may be
possible to isolate this highly con-
densed X (or Barr body) from inter-
phase cells, but this requires a highly
specific assay for X-chromosome DNA.
As a preliminary step we have ini-
tiated experiments designed to obtain
X-specific DNA sequences that may
provide the essential probe for the Barr
body isolation. We are attempting to
isolate X-specific DNA from the inac-
tive X-chromosome because the timing
of its replication enables us to label the
X-DNA selectively with bromo-
deoxyuridine (BrdU). The newly syn-
thesized DNA has a greater buoyant
density and can be separated on that
basis from other chromosomal DNA.
We have established short-term cul-
tures of leukocytes from the peripheral
blood of human females and exposed
these cells to a one-hour pulse of 3 H-
BrdU (S.A. 1 Ci/mM) during the termi-
nal period of DNA synthesis. After sev-
eral hours cells that had incorporated
BrdU into DNA reach metaphase and
are blocked at that stage with col-
chicine. The next step is to purify
chromosomes from interphase cells in
the culture. Chromosomes are sepa-
rated from contaminating interphase
nuclei by velocity sedimentation in a
sucrose gradient. The fractions con-
taining chromosomes — but no nu-
clei — are pooled, and the DNA is
extracted. Only late replicating DNA
incorporates the BrdU, thereby alter-
ing the buoyant density of DNA and
permitting the separation of the
BrdU-substituted DNA from the un-
substituted DNA by cesium chloride
density gradient centrifugation.
The BrdU-labeled DNA can be a
template for the synthesis of com-
plementary RNA and the cRNA ob-
tained can be hybridized in situ to
human metaphase chromosomes to de-
termine its specificity.
Examination of chromosomes from
cultured cells obtained just prior to cell
lysis has revealed that under the condi-
tions used, BrdU is incorporated into
the chromosomal DNA — and that most
of the label is confined to the
X-chromosome. However, under these
conditions, only a few metaphase cells
have labeled chromosomes. We have
been able to obtain a small smount of
BrdU-labeled DNA isolated as a satel-
lite of the main band (Fig. 8). The
quantity, however, has so far been in-
sufficient to use as template for cRNA
synthesis.
We are attempting to increase the
field of specifically labeled DNA and
decrease the small amount of contami-
nation released from breakdown of in-
terphase nuclei.
CHROMOSOMAL PROTEINS AND THE CONTROL OF GENE
FUNCTION
R. H. Reeder, A. Bokhon'ko, andH. L. Wahn
with the assistance ofE. Hogan
We have continued to improve the
method for isolating nucleoli from
Xenopus oocytes so that we can recover
not only the amplified ribosomal DNA
as chromatin but also several enzyma-
tic activities associated with it. We
have demonstrated both RNA poly-
merase I activity previously known to
be associated with rDNA as well as an
enzymatic activity that relaxes
superhelical turns in DNA. Some evi-
dence suggests this latter activity may
be involved in transcribing rDNA.
In Year Book 74, pp. 27-28, we pre-
sented evidence that at least some frac-
tion of the rDNA in Xenopus cultured
cells is in subunit configuration with
chromosomal proteins. This year we
expanded this observation by demon-
strating that the A+T-rich satellite
DNA of mouse is also organized with
proteins in the same subunit fashion.
24
CARNEGIE INSTITUTION
1.25
1.24
1.23
1.22 r
1.21
1.0
_
Nucleoli
2.0
1
1
h
Whole
i
'
Cells "
1.5 l
o
Nuclei
\
6
<
0.5
1 ' '\
/ ' • \
/ ' < \
\
0_
/; r
/ 1 *
o
1
I
/ 1 °
fO
/ ? .
y / / •
;_ /
0.5 £
o
E
a.
r\ m
a_S
•B-W" * 1
1
W*<o
u
5 10
15
20
Fract
ion
No.
nIOOO
<
- 500 or
k.
Q_
CVJ
ro
-10
Fig. 9. Isolation of amplified nucleoli on a metrizamide density gradient. A homogenate of immature
Xenopus oocytes was used to pour a metrizamide density gradient as described in the text. Fractions
were analyzed for light scattering. A 650 (• •); ribosomal DNA hybridization (• •); and RNA
polymerase activity (O O).
Isolation and Characterization
of Amplified Nucleoli
H. L. Wahn andR. H. Reeder
We have improved the previous
method for isolating amplified nucleoli
by the addition of a density gradient
separation as the final step. Immature
pigment-free oocytes were isolated es-
sentially as described in Year Book 73
(p. 28). The oocytes were then homo-
genized in a solution of metrizamide
and the homogenate used to pour a
linear metrizamide density gradient.
Metrizamide is an inert tri-iodinated
benzamido derivative of deoxyglucose
which can be used to form gradients
that cover the entire density range of
biological macromolecules. Since met-
rizamide is non-ionic, the gradients do
not dissociate nucleoli or other cellular
organelles and they can be isolated in
an apparently native state.
Figure 9 shows the result when a
metrizamide density gradient was used
to fractionate a whole oocyte homoge-
nate. Examination of each fraction by
phase contrast microscopy revealed
that nucleoli were banded at a rela-
tively high density location well sepa-
rated from contaminating nuclei.
Measurement of light scattering at 650
nm (metrizamide absorbs ultraviolet
light) showed a peak of scattering that
coincided with the location of nucleoli
in the gradient and which coincided
with a peak of rDNA as measured by
hybridization with radioactive rRNA.
If nucleoli were rebanded in a second-
round metrizamide gradient, it was
possible to reduce bulk chromosomal
DNA contamination to less than 10% of
the total DNA found in the nucleolar
peak (as judged by analytical ultracen-
trifugation analysis).
Nucleoli isolated by this method
have all the characteristics expected
for nucleoli that are active in rRNA
synthesis. As seen in Fig. 10 they show
RNA polymerase activity. This RNA
polymerase is sensitive to actinomycin
D and resistant to high concentrations
DEPARTMENT OF EMBRYOLOGY
25
of a-amanitin. Therefore, it has the
properties of the template requiring
polymerase I or nucleolar polymerase.
In addition, these isolated nucleoli ex-
hibit the normal morphology charac-
teristic of active nucleoli when
analyzed by electron microscopy, and
they contain both 40S and 30S rRNA
precursor.
However, when nucleoli are re-
banded on a shallow metrizamide gra-
dient, it is apparent that not all are
equally active in transcription. Figure
10 shows another gradient in which
RNA polymerase activity and rDNA
amount were measured in each frac-
tion. The more dense nucleoli have a
much higher polymerase to rDNA ratio
than do the less dense nucleoli. Al-
though we do not yet know the basis for
this subfraction of nucleoli, it has
1500-
1000-
m 500
CVJ
•^ 4
o
<
S 2
x
o
0)
a:
rDNA
/
\ Bulk DNA
\ 9 \
\ / o
V
Relaxing
Activity
Polymerase
- j! W*A
i/ \ \ /
or J
-•-U 1
1 1
- 500
.6,
<
rr
o
<
250 *->
3
m
o
V
i
0.3 ^
Q_
I-
-|0.2 V
X
5 10 15
Fraction no.
Fig. 10. Association of RNA polymerase-
relaxing and DNA-relaxing activities with nuc-
leoli. Nucleoli were banded in a second-round
metrizamide density gradient, and fractions
were assayed in various ways. Upper panel:
Ribosomal DNA hybridization (• •). Bulk
chromosomal DNA hybridization (o o).
Lower panel: DNA-relaxing activity (• •).
RNA polymerase activity (o o).
proved very useful in helping to deter-
mine which proteins are associated
with transcriptionally active rDNA.
Figure 11 shows an SDS-acrylamide
gel electrophoresis of total nucleolar
proteins from each fraction across a
second-round metrizamide gradient.
The distribution of RNA polymerase
activity and rDNA is also shown for the
same gradient. Our next major task is
to sort out these proteins, subtract
those which are involved only in ribo-
some assembly, and determine the
function and location of those proteins
which bind to the rDNA. Some pro-
teins, such as the four major histones,
have distributions that appear to coin-
cide with the distribution of rDNA.
Several other protein bands, however,
follow the distribution of RNA
polymerase activity and these are of
prime interest for future study.
One highly interesting protein that
distributes with the polymerase activ-
ity is an enzyme that relaxes super-
coiled DNA. To assay for this enzyme,
nucleolar fractions were incubated
with supercoiled Col El plasmid DNA
and the DNA was electrophoresed on a
1% agarose gel. A densitometer trace of
the stained gel is shown in Fig. 12.
Some of the supercoiled DNA has been
completely relaxed while some of it has
been trapped at intermediate levels of
supercoiling, giving rise to a series of
discretely spaced bands in the gel. The
relative level of DNA-relaxing enzyme
was assessed by measuring the amount
of completely supercoiled DNA left
after incubation for a standard time
interval. The distribution of relaxing
activity across a metrizamide gradient
is shown in Fig. 10. It is clear that the
relaxing enzyme tracks with RNA
polymerase activity and not with the
rDNA distribution. This suggests that
the relaxing enzyme may be an inte-
gral part of the transcription ma-
chinery. It seems unlikely that the pre-
sence of this relaxing enzyme is due to
the presence of DNA replication, since
these nucleoli are in a replication dead
end. They have previously replicated
their DNA weeks to months before iso-
26
CARNEGIE INSTITUTION
H 123456789
-130
- 68
- 60
-43
25.7
H2b
H3
*
H2a
H4- ■
- 17.2
- 11.7
1500-
3 4 5 6 7 8 9
Fraction number
Fig. 11. SDS-acrylamide gel electrophoresis of proteins from nucleoli banded in metrizamide. Nuc-
leoli were banded in a second-round metrizamide gradient, and protein from each fraction was elec-
trophoresed on a 15% acrylamide-SDS gel. Aliquots of each fraction were also assayed for RNA
polymerase activity (o o) and ribosomal DNA location (• •). Lane H shows acid-extracted
Xenopus histones. Molecular weights are in daltons x 10~ 3 .
lation and they will never enter the re-
plication cycle again. Furthermore, we
cannot detect any significant DNA
polymerase activity associated with
these nucleoli even after adding an
exogenous template.
Since amplified rDNA is most likely
present as circular molecules, it is
quite reasonable that a relaxing en-
zyme might be necessary for some
phase of transcription. Initiation by
RNA polymerase requires localized
DEPARTMENT OF EMBRYOLOGY
27
Fig. 12. Assay of nucleolar DNA relaxing en-
zyme. Supercoiled ColEl plasmid DNA was in-
cubated in the presence (upper trace) and in the
absence (lower trace) of nucleoli purified through
metrizamide. The ColEl DNA was then elec-
trophoresed on a 1% agarose gel, and the DNA
was located by staining with ethidium bromide.
Direction of electrophoresis is from right to left.
Supercoiled ColEl (major peak in lower trace)
has a higher mobility than relaxed circular
ColEl DNA (major peak in upper trace). The in-
termediate bands differ from each other by inte-
grals of one supercoil turn.
strand separation of the DNA and
therefore must introduce superhelical
strain into the DNA. It is also possible
that chain elongation causes similar
physical changes in DNA structure,
especially if the polymerase must dis-
lodge histones as it transcribes the
gene. Therefore, an enzymatic activity
that would release physical strains
placed on the DNA by transcriptional
processes would be required to return
the DNA to its normal state.
The Subunit Organization of
Mouse Satellite Chromatin
A. Bokhon'ko
Most of the DNA in a eukaryotic
chromosome is complexed with his-
tones to form subunits resembling
beads on a string. This structure can be
observed with the electron microscope
and can be demonstrated in several in-
direct ways. One property of this sub-
unit organization is that DNA con-
tained within the bead is more resis-
tant to endonuclease attack than is
DNA between the beads. Thus, limited
digestion of chromosomes with nu-
clease cleaves the DNA into small
fragments that are multimers of about
200 base pairs of DNA. An example of
this phenomenon is shown in Fig. 13.
In this experiment mouse liver nuclei
were briefly digested with nuclease,
and the DNA was deproteinized and
electrophoresed on an acrylamide gel.
An oligomeric series of bands is ob-
served, resulting from histone protec-
tion in the chromatin subunits.
It is of interest to know if there are
any specific DNA sequences that are
exempt from this type of subunit or-
ganization. Last year we presented
evidence that some fraction of the
ribosomal genes were so organized and
suggested that histones may be present
on actively transcribed sequences. This
year we have used similar techniques
to examine the A+T-rich satellite se-
quences in mouse DNA. This DNA is
located in highly condensed cen-
tromeric heterochromatin and appears
never to be transcribed in the cell.
The experiment is shown in Fig. 13.
After brief digestion of mouse liver
nuclei with micrococcal nuclease the
protected DNA fragments were elec-
trophoresed on an acrylamide gel. The
gel was then sliced and the DNA in
each slice was hybridized with radioac-
tive RNA transcribed in vitro from
purified mouse satellite L-strand. The
hybridization essentially follows the
distribution of the bulk DNA, and we
conclude that mouse satellite is also
organized in a similar subunit fashion.
Together, these results increase the
likelihood that histones are associated
with all of the chromosomal DNA re-
gardless of its function.
28
CARNEGIE INSTITUTION
SLICE NO.
Fig. 13. Hybridization of mouse satellite cRNA to micrococcal nuclease fragments from mouse liver
nuclei. Mouse liver nuclei were partially digested with micrococcal nuclease, and the DNA fragments
were purified and electrophoresed on a 6% acrylamide gel. The gel was sliced and the DNA in each
slice was hybridized with :! H cRNA transcribed in vitro from the L-strand of purified mouse satellite
DNA. The bulk DNA was located by staining with ethidium bromide, (a) Fragments from nuclei
digested for 5 min. (b) Fragments from nuclei digested for 20 min. The arrow indicates the position of
the presumed monomer at approximately 170 base pairs.
THE POLARITY OF RIBOSOMAL DNA IN Xenopus
I. B. Dawid, R. H. Reeder, P. K. Wellauer, T. Higashinakagawa
in collaboration with O. L. Miller, Jr.*
Ribosomal RNA (rRNA) in Xenopus
and other eukaryotic nuclei is trans-
cribed as a precursor molecule contain-
ing both 18S and 28S RNA regions and
transcribed spacer segments. The ques-
tion of whether the 18S or the 28S re-
gion is closer to the 5' end of the pre-
cursor has been answered in opposing
ways (Year Book 69, 565; Year Book 72,
38; Year Book 73, 45; and Year Book
74, 20). We have undertaken a re-
investigation of this question which
leads to the strongly supported conclu-
sion that the 18S RNA sequence is
closer to the 5' end (transcription ori-
gin) of the precursor than is the 28S
*University of Virginia.
RNA sequence. This order of sequences
is almost certainly correct for rRNA
precursor molecules of other
eukaryotes as well. Transcription in
the direction of small-to-large rRNA is
a general feature of genomes, including
prokaryotes, eukaryotic nuclei,
chloroplasts and mitochondria. A
model of a segment of Xenopus laeuis
ribosomal DNA with the correct polar-
ity of the precursor is shown in Fig. 14.
Determination of Polarity with
DNA Exonucleases
The first test of polarity in rRNA
that we describe depends on the
DEPARTMENT OF EMBRYOLOGY
5'
29
3'
5'
//-
40S -*- I //
I8S 28S NTSp I8S
Xlr
3 5
H h
//
5" 3' Xlr 12
X- EXONUCLEASE
18 S
28 S
EXONUCLEASE IE
28 S
I8S
Fig. 14. Determination of rDNA polarity with DNA exonucleases. The figure is described in the
text.
availability of fragments of rDNA
whose ends have been mapped with re-
spect to the rRNA (see Fig. 14) and of
DNA exonucleases with known speci-
ficities. Exonuclease III from E. coli
(specific for 3' ends) and lambda
exonuclease (specific for 5' ends) were
donated by former fellows of our De-
partment, John Chase and John Mor-
row, respectively. The principle of the
experiment is illustrated in Fig. 14. At
the top we show a model of a segment of
rDNA with the polarity that is sup-
ported by the experiments reported
here. EcoRI cleaves each repeat twice,
creating two types of fragments. Such
fragments have been cloned in E. coli
and two particular ones are named
Xlr 11 and Xlrl2. In Fig. 14 the coding
strand is the "heavier" line and the po-
larity of each strand is indicated in
agreement with the polarity of the
RNA. Such rDNA fragments were par-
tially digested with each of the two
exonucleases until about 20% of the
DNA was hydrolyzed. The models in
Fig. 14 illustrate the resulting
molecules. Lambda exonuclease diges-
tion of Xlrll removes the anti-coding
strand in the 18S region, thereby ex-
posing 18S sequences on the coding
strand. At the other end of Xlrll the
same exonuclease digests the coding
strand only. The resulting DNA frag-
ment with single-stranded tails is
therefore expected to hybridize with
18S rRNA only, as indicated in the fig-
ure by dotted line. Similar models for
the other fragments and both exonuc-
leases are shown in Fig. 14. When the
opposite polarity is assumed in the
rRNA, the predictions for exonuclease
III and lambda exonuclease are simply
interchanged. We consider the con-
clusion from these experiments to be
firm because the specificities of the two
exonucleases have been well estab-
lished and because the test involves
four separate experiments (two frag-
ments, two exonucleases) all of which
give highly consistent data in support
of the same polarity.
Determination of Polarity
Using rDNA Transcription
Complexes
The second test of polarity makes use
of the fact that ribosomal gene trans-
30
CARNEGIE INSTITUTION
cription complexes can be spread for
viewing in the electron microscope.
This technique was first developed by
O. L. Miller, Jr., at the University of
Virginia, and this experiment was
done with his collaboration. The nas-
cent rRNA chains in such transcription
complexes are seen to be short at one
end of the complex (the 5' end of the
precursor) and to increase in length
until they are released at the other (3')
end. To determine which end the 18S
rRNA sequence was near, we digested
transcription complexes with the re-
striction enzyme EcoRI which cuts
rDNA at two known locations within
the transcribed region.
The cut transcription complexes
were then spread for electron micros-
copy, and we measured the distances
from cut sites to either termination (3'
end) or initiation (5' end) sites.
The experiment is diagrammed in
Fig. 15. In panel (a) a full-length trans-
cription complex is shown at top with
the location of EcoRI sites indicated,
assuming the 18S sequence is proximal
to the 5' end of the precursor. Below
are shown the four types of transcrip-
tion complex fragments that would re-
sult from this model. Four types of cut
ends are generated because proteins on
the rDNA block the EcoRI sites with a
certain probability so that only partial
digestion is achieved. Panel (b) shows
the predicted fragments if the opposite
polarity is assumed.
Measurement of 100 EcoRI cut
transcription complexes yielded results
that agreed well with model (a) and
strongly disagreed with model (b). This
supports the conclusion that the 18S
sequence is proximal to the 5', or initi-
ation, end of the rRNA precursor.
a) 5'-l8S
* «H .„„ I — h
I8S 28S
t t
EcoRI EcoRI
b) 5'-28S
I I ,„„ I — ' — f
28S ' ' I8S
t t
EcoRI EcoRI
.33
.93
.07
.67
.67
.33
.93
Fig. 15. Determination of rDNA polarity using EcoRI on transcription complexes. The figure is
described in the text.
RIBOSOMAL DNA IN Drosophila melanogaster
P. K. Wellauer and I. B. Dawid
in collaboration with K. D. Tartof of the Institute for Cancer Research Philadelphia,
and with the assistance ofM. Rebbert
The genes that code for the large occur in multiple copies. The ribosomal
ribosomal RNA (rRNA) molecules DNA (rDNA) forms clusters of redun-
DEPARTMENT OF EMBRYOLOGY
31
dant genes in the nuclei of all or-
ganisms studied. Since the rDNA of
many species has special physical
properties, it may be isolated and
characterized. The rDNA of Xenopus
laevis was the first eukaryotic gene to
be purified (Year Book 67, 401), and
this DNA has since been characterized
extensively (see Year Book 74, p. 19,
and this Year Book). In spite of the fact
that Xenopus rDNA had been isolated
some time ago and the isolation of
rDNA from other organisms seemed
technically feasible, few reports on the
purification and detailed characteriza-
tion of rDNAs have appeared in the lit-
erature. It is clear that the properties
of rDNA are of great interest, and that
generalizations about the structure
and evolution of this DNA can be based
only on the comparison of results with
several organisms. We decided to study
rDNA in Drosophila melanogaster be-
cause this organism offers the advan-
tage of producing many known mu-
tants that affect the ribosomal locus,
and because it shows the interesting
phenomena of compensation and mag-
nification of rDNA.
Purification of Drosophila rDNA
In constrast to X. laevis, Drosophila
rDNA is not a density satellite, i.e., the
buoyant density of rDNA is very close
to the density of the average DNA. We
achieved purification of the rDNA in
two different ways. In the first proce-
dure Drosophila DNA, purified from
embryos, was centrifuged in cesium
sulfate in the presence of 0.2 moles of
mercury ions per mole of nucleotide.
The rDNA was located by hybridiza-
tion with rRNA, the corresponding re-
gion of the gradient pooled, and the
DNA centrifuged in cesium chloride in
the presence of actinomycin D. Again,
rDNA was pooled and centrifuged in
cesium sulfate with mercury at a ratio
of 0.08. The rDNA pooled from this
third gradient was usually "pure,"
which means that we estimate its pur-
ity above 80% (see below).
The second method for rDNA purifi-
cation is based on the R-loop procedure
developed by White and Hogness of
Stanford University. When native
DNA and RNA are annealed in high
concentrations of formamide at the
right temperature, the RNA invades
the duplex and displaces one of the
DNA strands. The resulting complex is
suitable for mapping by electron mi-
croscopy and also for preparative pur-
poses, since the RNA/DNA hybrid is
more dense than native DNA. Thus,
Drosophila DNA was incubated with
rRNA and centrifuged in cesium sul-
fate. The rDNA region, which was
slightly more dense than the main
band of DNA, was pooled and cen-
trifuged again in the same way. The
r
Fig. 16. Electrophoretic patterns of Drosophila
rDNA digested with the restriction endonuclease
Smal. Electrophoresis is from top to bottom; the
gels were stained with ethidium bromide and
photographed in ultraviolet light. Lane a,
marker fragments from a Hindlll digest of
lambda DNA; lane b, partially purified rDNA
containing 10-20% rDNA; lanec, rDNA purified
by the mercury-actinomyein D method.
32
CARNEGIE INSTITUTION
resulting rDNA was pooled and the
RNA removed. The product was effec-
tively pure rDNA of high molecular
weight.
The purity of rDNA is most easily
checked by digestion with the restric-
tion endonuclease Smal. Figure 16
shows that rDNA is cleaved into one
class of fragments of about 8.5 kb
(kilobases or kilobase pairs); this class
is heterogeneous because of length
heterogeneity in the rDNA repeating
units (see below). Further, there are
discrete fragments of 2.8, 2.3 and 2.2
kb. Since Smal sites are rare in main-
band DNA, contaminating DNA is
largely uncleaved and appears in the
gel as a broad band moving more
slowly than the 8.5-kb fragment set.
This is easily seen in Fig. 166, which
shows the pattern given by an Smal di-
gest of partially purified rDNA. By
comparison, lane c of Fig. 16 shows lit-
tle if any contaminating material.
Analysis of EcoRI Fragments of
Drosophila rDNA
The restriction endonuclease EcoRI
cleaves rDNA into a number of frag-
ments (Fig. 17a). Predominant bands of
16.5 and 11 kb are seen. In addition,
there are bands of intermediate size,
STAIN
rRNA
X
Y
I #H KB
Fig. 17. Electrophoretic patterns of Drosophila rDNA digested with the restriction endonuclease
EcoRI. Lane a, from wild-type male and female flies stained with ethidium bromide. Lane b, purified
wild-type rDNA was digested and separated by electrophoresis, and the DNA fragments were trans-
ferred to membrane filters as described in Southern (Journal of Molecular Biology 98, 503-517, 1975),
and hybridized with labeled rRNA. The figure shows the autoradiogram of the hybridized filter.
Lanes c and d, autoradiograms of rDNA patterns from DNA isolated from flies that carried a bobbed
mutation on their Y chromosome so that at least 95% of the rDNA is derived from the X chromosome.
Lanee, rDNA pattern from flies of the genotype sc 4 -sc K /Y in which the nucleolus organizer is deleted
from the X chromosome.
DEPARTMENT OF EMBRYOLOGY
33
and many bands smaller than 11 kb,
among which two bands at 7.4 and 5.4
kb are relatively abundant.
Electron microscopy was used to
analyze the structure of EcoRI frag-
ments of rDNA. We first summarize
the results and then illustrate some of
the evidence for it. The fragments of 1 1
kb and larger are the results of a single
EcoRI cut in the 18S RNA region of
each rDNA repeating unit. The 11-kb
fragments have a structure similar to
that of Xenopus rDNA: The 18S and
28S rRNA regions are separated by a
short spacer segment that is likely to
be transcribed (Spl in Fig. 18B) and by
a longer spacer segment that must con-
tain mostly nontranscribed sequences
(Sp2 in Fig. 18B).
The 16.5-kb fragments contain a
rather unexpected type of structure: In
addition to the regions found in the
11-kb fragments there is a spacer seg-
ment that is inserted within the 28S
gene region. We call this spacer the
"insertion." Figure 18A shows that this
insertion is located 1.13 kb from the
end of the 28S gene. In Drosophila as in
other insects, 28S rRNA as isolated has
a break about in the middle. The posi-
tion of this break is seen in the maps of
Fig. 18 as a gap, suggesting that the
break is generated from the original
continuous transcript by the removal of
about 140 bases. It is important to note
that the position of the spacer insertion
and the position of the gap are sepa-
rated by 740 base pairs. The insertion
occurs in different size classes. The
most abundant class is a long insertion
that is present in the 16.5-kb type of
repeating unit. Other insertion sizes
account for the EcoRI fragments found
in the gel between the 11- and 16.5-kb
positions. Remarkably, the insertions
occur in size classes that are multiples
of a basic size of close to 0.5 kb.
Whether the larger insertions are ac-
tually repetitions of the 0.5-kb se-
quence is not known. EcoRI fragments
smaller than 11 kb are produced by ad-
ditional cuts in some repeating units of
rDNA.
Before we discuss the implications of
these structures we show some of the
evidence for them in Fig. 19. The figure
shows hybrid molecules between single
strands of rDNA EcoRI fragments and
rRNA. In the DNA fragments that con-
tain insertions, one 28S rRNA
molecule hybridizes with its com-
plementary sequences on both sides of
Gap
I8S Spl
a
28Sb
0.20 1.23
■ V////////;/7 )r V/////////77777l -
! 1.87
1.59
Sp 2
I8S
1.56
Ins 0.5 - 6.0
28Sb' 28Sb"
0.74
1.13
-ZZZZZZZZZZZA
Fig. 18. Models of EcoRI fragments of Drosophila rDNA. These fragments constitute complete
repeats. The EcoRI site is in the 18S RNA region. Spl is likely to be transcribed; Sp2 is mostly
nontranscribed. The gap in the 28S RNA region is the position where an interruption occurs in the
28S RNA. Sp2 is shown as a broken line, since length heterogeneity occurs in this region. Model B
shows the 11-kb fragment; the numbers are the sizes of the different regions in kb. Model A shows the
class of fragments that contain an insertion of spacer sequences within the 28S gene region. Those
regions not labeled in model A have the same length as in model B. The insertion occurs in size classes
from 0.5 to 6.0 kb and in multiples of 0.5 kb.
34
CARNEGIE INSTITUTION
H^j[. ' ■'•,-■ ;^;\ *:-' ; v^
■■••■■- •-,• ■ >.-•••.. -, •■• • ■■. .-r— -*-. ■■*■-•• ', -¥£/ H l l >. '•;->;■ ••■>■•
"TV.
' s P 2 -•"-; H : ^, "M-
;•'?-'
' .:'V'<v^'" \ '•'•"•
• : v ' "-'i-f
>-.■••?••;,•'•; '■■s^SvfrZ. -v. * --^A^/} ■ •.■■^■:^.i> remiss* i,>
ge-v-
: > :.:'/■ : - ■■'■■*■:
M
Sp2
:"-. n**"-->' "^ " / '■* •• -f -.■" "■ I - w ' ■ ■'( V 1 '- ■"»'' '*•'' ^' v '■^'1 • ""'~'' ' ■ ll '''' S' " r -•^^* , *fc-tl*->
;<i8s;+s P i • .• ^ .''"-/?; vL ' s ^* 1 .'.'- > ' < CJj. " -■ - •
Fig. 19. Electron micrographs of hybrid molecules between EcoRI fragments of Drosophila rDNA
and rRNA. Hybrids were prepared for electron microscopy by the formamide modification of the
Kleinschmidt technique. The various regions are indicated and are explained in the legend to Fig. 18.
Insertions are indicated by arrows. (A) 11-kb EcoRI fragment. (B) EcoRI fragment with a 1.5-kb
insertion. (C) EcoRI fragment with a 5.0-kb insertion.
DEPARTMENT OF EMBRYOLOGY
35
the insertion so that the insertion
forms a loop. Such molecules, which
were frequent, show most clearly that
the position of the strand interruption
in the 28S RNA and the position of the
insertion in the gene are distinct. In
Fig. 19 we show one 11-kb repeat and
two insertion repeats, one from the
major (16.5-kb) class and another
molecule with a 12.5-kb repeat that
has a 1.5-kb insertion.
These observations raise the ques-
tion of whether, and if so how, the
rDNA repeats with insertions are
transcribed. To produce intact 28S
rRNA there would have to be trans-
cription from gene sequences on both
sides of the insertion with subsequent
ligation of the products by a hypotheti-
cal RNA ligase. Or, the RNA
polymerase would "jump" across a loop
of DNA, as the RNA does in the hy-
brids of Fig. 19. Such a mechanism has
not been described in any system. Al-
ternatively, the genes with insertions
may not be transcribed, or transcrip-
tion may end at the insertion point (we
assume that transcription begins to the
left of the 18S region and proceeds to-
wards the 28S region. In the latter case
RNA transcripts would arise from the
insertion-carrying genes, but they
could not produce complete 28S rRNA
and would be nonfunctional. One
would then ask why the inserted genes
have been conserved during evolution.
At present we cannot choose among
these alternatives and cannot answer
the questions raised by the observa-
tions.
Length Heterogeneity in the
Spacer Region
In addition to the length differences
in the insertion there is also length
heterogeneity in spacer 2 (Fig. 18),
which corresponds to the "standard"
nontranscribed spacer of X. laevis
rDNA. This length heterogeneity is
'less extreme than in Xenopus, with
most spacers falling into a fairly nar-
row distribution. Nevertheless, we
have obtained unequivocal evidence for
the existence of heterogeneity by
examination of heteroduplex
molecules. EcoRI fragments of rDNA
were separated on gels, and 11-kb,
16.5-kb, and intermediate fragments
were isolated. These fragments were
denatured and allowed to reassociate.
Many of the duplex molecules that
formed had single-stranded deletion
loops in the spacer region. Especially in
those molecules where only a single
loop appeared we may conclude that
the spacers of the two strands which
formed the duplex differ in length.
Drosophila rDNA has two regions in
which length heterogeneity occurs: the
insertion, which comes in multiples of
0.5 kb and accounts for most of the
heterogeneity; and the nontranscribed
spacer, which varies in length in a nar-
row range without detectable size clas-
ses.
Arrangement of Length
Heterogeneity
We have tried to resolve whether
rDNA repeats without insertions and
with insertions of different length
occur in clusters, or whether they are
interspersed along the chromosome.
High molecular weight rDNA was iso-
lated, denatured, hybridized with
rRNA, and inspected in the electron
microscope. The results demonstrate
that repeats of different length are in-
terspersed. Repeats of 11 kb may occur
next to 11-kb, 16.5-kb, or intermediate
length repeats, and the same is true for
all classes of repeats. A statistical
evaluation of 85 pairs of adjacent re-
peats shows that the frequencies with
which different types of repeats are
nearest neighbors are close to those ex-
pected from a random assortment of
the different length classes. In some
cases a tendency toward clustering of
similar repeats was seen, and we can-
not decide whether the distribution is
fully random or not. But it is clear that
extensive scrambling of different re-
peat sizes occurs in Drosophila rDNA,
36
as it does in the rDNA of Xenopus
{Year Book 74, p. 22).
Distribution of rDNA Repeat
Lengths on the X and Y
Chromosomes of D. Melanogaster
\nD. melanogaster the rDNA is clus-
tered in one nucleolus organizer each
on the X and the Y chromosome. About
200 repeats of rDNA compose each wild
type nucleolus organizer. To test the
structural properties of rDNA derived
from the two genetic loci we digested
DNA from the appropriate stocks with
EcoRI, separated the fragments by
electrophoresis, and hybridized the
fragments with rRNA by the method of
Southern. Figure 17 shows some of the
results. While in resolution the hy-
bridized gels do not match the stained
gel of purified wild type rDNA, it is
clear that the X chromosome carries
both major classes of repeat length (11
kb and 16.5 kb), while the Y chromo-
some carries the 11-kb fragment but
not the 16.5-kb fragment. Minor frag-
ments occur on both chromosomes but
are not well displayed in Fig. 17 c to e.
These observations raise some intri-
guing questions about the evolution of
tandem repetitive gene families and
about the "correction mechanism" pos-
carnegie institution
tulated to operate during this evolution
(see Year Book 74, p. 19). There is prior
genetic evidence against recombina-
tion between the X and the Y chromo-
somes. The absence of 16.5-kb frag-
ments on the Y chromosome supports
this evidence: If such recombination
occurred, one would expect the 16.5-kb
fragments to spread to the Y chromo-
some unless some unknown mecha-
nism prevents these repeats from oc-
curring on the Y chromosome. If, how-
ever, the X and the Y are genetically
isolated, how are the similarities be-
tween them maintained? These
similarities are extensive, since rRNA
transcribed from either chromosome is
very similar and probably identical,
and the 11-kb fragments and some
other features of gene organization are
closely similar in X and Y rDNA. If
there were no genetic exchange, one
would have to postulate that selective
pressures maintain all these features
on the two separate chromosomes. It
appears to be difficult to adhere to such
a hypothesis in its strict form. How-
ever, we would like to raise the possi-
bility that selection plays a larger part
in the maintenance of homogeneity in
tandem gene families than had been
believed. Drosophila rDNA seems
highly suited for experiments that will
aid in understanding these processes.
STRUCTURAL AND FUNCTIONAL STUDIES OF FIBROIN GENES
Y. Suzuki, Y. Ohshima, P. Geshelin, and P. E. Giza
We wish to know how a tissue-
specific gene is transcribed in that tis-
sue at a given time. More practically,
our long-term goal is to understand fi-
broin gene control by reproducing its
function in vitro with the purified gene
and other necessary components. Each
gene system has certain advantages
and disadvantages for this kind of
study. The fibroin gene system in the
posterior silk gland of Bombyx mori
has the advantages that (1) the coding
sequence is extremely simple, inter-
nally repetitious, and of high G + C con-
tent; (2) the gene comprises a high pro-
portion of the genome; (3) its expres-
sion is temporally regulated, and fi-
broin mRNA comprises an unusually
high proportion of cellular RNA at cer-
tain developmental stages; and (4) iso-
lated nuclei also synthesize the mRNA
in high proportion. Making use of these
advantages, we have been employing
three approaches toward our goal: (1)
development of methods for purifying
fibroin genes, (2) development of
methods for cloning genes, and (3)
transcriptional analysis of fibroin
DEPARTMENT OF EMBRYOLOGY
37
genes in vivo and in vitro. Some of our
progress is summarized below.
Physical Purification of the
Gene Coding for Silk Fibroin
P. Geshelin
In order to characterize the manner
in which silk fibroin synthesis is con-
trolled, we need to isolate the DNA
that codes for the protein. Single-copy
genes from various organisms are
being isolated in several laboratories
at present using one of two general
methods. In the first, a small amount of
whole-organism DNA is partially
purified by proved physical methods,
enzymatically attached to plasmid
DNA, and grown in infected bacteria.
After the clone(s) of bacteria that con-
tain the sought-after sequences are
selected (usually by hybridization to
the corresponding messenger RNA),
the gene is replicated many times over
by growth of the clone(s). In the second
method the tiny amount of desired
DNA is isolated physically from a huge
amount of DNA from the whole or-
ganism by hybridizing messenger RNA
to the sheared DNA and separating the
DNA-RNA hybrids from the rest of the
DNA by utilizing the physical prop-
erties of the RNA. One makes use of
the higher density of the RNA, com-
pared to DNA. This density differential
can be increased by binding substances
to the RNA before hybridization. Al-
ternatively, the RNA can be covalently
bound to a solid support (i.e., agarose
beads) before hybridization, and the
DNA not annealed to it after hybridi-
zation may be simply washed away.
Attempts are under way to isolate
silk fibroin DNA by a modification of
these physical methods. Messenger
RNA coding for silk fibroin will be co-
valently bound to mercury atoms by
procedures developed by Ward and
co-workers (Biochemistry 14, 2447-
2457 and 2458-2469, 1975). The mes-
senger RNA will then be mixed with
whole-cell DNA and placed at the melt-
ing temperature of the DNA in high
formamide. Because RNA-DNA dup-
lices are more stable than DNA-DNA,
especially in the solvent formamide,
the RNA displaces the minus or coding
strand of DNA from the helix over the
region to which the RNA is homolo-
gous, forming a triple-stranded struc-
ture termed an R-loop. These R-loops
will then be separated from the rest of
the nucleic acid by affinity chromatog-
raphy on sulhydryl agarose, which
selectively binds the mercury atoms.
This new method could have several
advantages over previous procedures
for purifying single-copy genes. First,
one will get both strands of the silk fi-
broin DNA, in contrast to the methods
involving melting and annealing,
which yield only the coding strand.
Second, affinity chromatography
theoretically offers much greater reso-
lution between mercurated and non-
mercurated nucleic acids than does
density centrifugation between DNA
and RNA. Finally, since the resolution
of the separation technique does not
depend on DNA size, one can strive to
isolate whole-gene DNA with large
flanking sequences thought to contain
important control sequences.
Cloning of Fibroin Gene Plasmid
Y. Ohshirna and Y. Suzuki
The fibroin gene exists as a single
copy per haploid complement. How-
ever, the gene is so large (about 12-14
x 10 6 daltons) that it comprises 0.004%
of the B. mori genome. In order to be
able to carry out the structural and
functional analyses of the gene, we
need to have large amounts of it in a
pure form. Therefore, we are trying to
obtain an E. coli clone that carries a
recombinant plasmid including the
whole fibroin gene with some adjacent
DNA. Such plasmid DNA, when it is
isolated, can be produced in milligram
quantities and will be of crucial impor-
tance for our experiments.
The plasmid we have chosen is
tetracycline-resistant Col El type
plasmid PMB9. It has a molecular
38
CARNEGIE INSTITUTION
weight of 3.5 x 10 6 and carries a repli-
cation function in the presence of
chloramphenicol and one restriction
site each for the endonucleases EcoRI
and Hindlll. In order to recombine this
plasmid with a large DNA molecule
containing the full-size fibroin gene
and some extra sequences (about 25 x
10 H daltons) we employ the dA-dT join-
ing method. Covalently closed DNA of
PMB9 was isolated and made into
linear forms by EcoRI. For poly(dA) or
poly(dT) addition we use terminal
deoxynucleotidyl-transferase from calf
thymus in the presence of 1 mM CoCl 2
(in either the presence or the absence of
MgCl 2 ). We have succeeded in adding
the single-stranded tail of poly(dA) or
poly(dT) directly to EcoRI ends of the
plasmid DNA without pretreatment of
the DNA by lambda exonuclease. Most
of the dA or dT incorporation was found
co-sedimenting with the plasmid DNA
in a neutral or an alkaline sucrose gra-
dient. When observed under the elec-
tron microscope, 80-90% of the ends of
the plasmid DNA have a single-
stranded tail.
For the fibroin gene part, in order to
obtain the adjacent control regions as
well as the whole coding sequence in a
plasmid we must start with a high
molecular weight preparation of B.
mori DNA, since the gene is extraordi-
narily large. We isolated a large quan-
tity of DNA with a molecular weight of
15-30 x 10 6 from frozen silk glands.
After two successive CsCl-gradient
bandings in the presence of actinomy-
cin D, we obtained a DNA preparation
in which fibroin gene DNA assayed by
hybridization with 125 I-mRNA was en-
riched by 40-fold (the gene concentra-
tion is about 0.2%). With this DNA we
propose to make poly(dT) tail additions
to the poly(dA)-tailed PMB9 DNA fol-
lowed by transformation of E. coli. E.
coli clones carrying the fibroin gene-
PMB9 plasmid will be selected by tet-
racycline resistance and hybridizabil-
ity with 125 I-mRNA of high specific ac-
tivity (about 1 x 10 4 cpm/ng.
Accentuated Expression of
Fibroin Genes in vivo
and in vitro
Y. Suzuki and P. E. Giza
We have developed a direct chemical
method for quantitating fibroin mRNA
in RNA mixtures {Year Book 73, p. 20).
Using this method we have measured
the amount of fibroin mRNA in the
posterior silk gland of B. mori. Fibroin
mRNA accumulates throughout the
fifth instar, ultimately comprising as
much as 4.4% of the total cellular RNA
toward the end of the instar. The
number of fibroin mRNA molecules ac-
cumulated by the end of the fifth instar
was 5.8 x 10 4 molecules/gene, about
5-fold higher than the number of rRNA
molecules (1.1 x 10 4 molecules/gene).
This observation has led us to estimate
the rates of fibroin mRNA and rRNA
accumulation at various stages of the
instar. As shown in Table 2, it was
found that (1) the rate of mRNA ac-
cumulation is almost constant
throughout the stage, whereas that of
rRNA changes as much as 30-fold, and
(2) mRNA accumulation relative to
gene number is proceeding at a higher
TABLE 2. Rates of Accumulation of Fibroin
mRNA and rRNA during the Fifth Instar
rRNA
Fibroin mRNA
Time Intervals*
molecules/gene/min
Ecdysis - 1st day
1st day - 2nd day
2nd day - 4th day
4th day - 6th day
6th day - 7th day
1.8
6.6
2.6
0.3
0.2
< 1.2
10
7.9
7.4
9.3
* Every 24-hr period after the fourth ecdysis
was designated as 1st day, 2nd day, etc. The
quantitation of the RNAs was done at ecdysis
and at the end of 1st, 2nd, 4th, 6th and 7th days.
The difference in amount of RNA between two
time points was divided by the total number of
genes and the time elapsed (24 hrs or 48 hrs in
minutes). The number of genes was calculated
from the amount of DNA present at the middle of
the two time points, with the further assumption
that each haploid complement of DNA contains
240 rRNA genes and 1 fibroin gene.
DEPARTMENT OF EMBRYOLOGY
39
rate than that of rRNA not only at
later stages of the instar but through-
out the instar except on the 1st day.
The most exaggerated situation is seen
on the 7th day of the instar. At this
stage new rRNA synthesis almost
ceases while mRNA accumulation con-
tinues actively. These results as well as
the consideration of the high propor-
tion of fibroin genes in the genome
have prompted us to try to detect fib-
roin mRNA in pulse-labeled RNAs by
our direct chemical method.
The calculated proportion of fibroin
mRNA synthesized and accumulated
for different labeling periods of 10 to
720 min at various stages of the fifth
instar and the preceding fourth moult-
ing stage are summarized in Table 3.
As expected from the exaggerated situ-
ation found on the 7th day of the instar,
fibroin mRNA was detectable even by a
10-min pulse, and occupied as much as
about 4% and 6% of the total cellular
RNA pulse-labeled on the 7th and 8th
day of the instar, respectively (Table
3). These are the highest proportions of
transcription ever found for a single
mRNA species. The proportion of
mRNA becomes even greater with
longer pulses, and reaches as high as
24% of the total RNA labeled in a
240-min pulse, indicating that fibroin
mRNA is much more stable than other
heterogeneous RNAs. When 30-min
pulses were given at various stages of
the fifth instar as well as the fourth
moulting stage, it was found that fi-
broin mRNA was not detectable during
the moulting stage, but in contrast, as
the fifth instar proceeded the propor-
tion of mRNA increased from 1.2% up
to 7.5% (Table 3). These observations
indicate that the synthesis of RNAs of
non-fibroin mRNA type is active at the
earlier stages, decreasing drastically
toward the end of instar, since it was
already shown that the rate of fibroin
mRNA accumulation is constant
throughout the stage (Table 2). As for
the moulting stage, we cannot deter-
mine whether fibroin genes are inac-
tive or whether their transcription is
not detectable simply because of a
rapid and selective degradation of fi-
broin mRNA.
From the proportion of fibroin
mRNA (Table 3) and rRNA (not shown)
synthesized in a 4-hr and a 12-hr pulse
in the 7th day we calculated rates of
each RNA synthesis per gene. The
ratio of the rates of fibroin mRNA syn-
thesis to that of rRNA synthesis was
about 50 to 1 both for the 4-hr and
12-hr pulses. This result indicates that
fibroin mRNA is being synthesized
TABLE 3. Proportion of Fibroin mRNA to Total Cellular RNA
i Synthesized during Short-Pulse Periods*
Time of Pulse Labeling
10 min
30 min
90 min 240 min
720 min
Staget
Proportion (%)
Fourth moulting
...*
<1.7
<0.5
Fifth instar
1st day
0.6
2nd day
1.2
1.8
3rd day
2.0
4th day
3.0
4.6
6th day
4.8
10
7th day
3.9
7.5
12 24
27
9th day
5.9
* Pulse-labeling experiments were done in the middle of the day.
t Defined as in Table 1.
tNot determined.
40
CARNEGIE INSTITUTION
about 50 times more rapidly than
rRNA. Because the ratios are the same
between the two time points, we infer
that rRNA and fibroin mRNA have
similar stabilities during these label-
ing periods and that the ratio of trans-
cription for these genes would also be
near 50 during shorter labeling periods
at this stage.
Because of this accentuated expres-
sion of fibroin genes in vivo, we have
tried to isolate and characterize silk
gland nuclei aiming at transcription in
vitro (Year Book 74, p. 36). The nuclear
isolation method has been applied to
the determination of the distribution of
unlabeled and newly synthesized fi-
broin mRNA in the cell. The results are
shown in Table 4. More than 80% of the
long lived mRNA which was identified
by A 2 6 o oligonucleotide profile after
T , RNase digestion was found in the
cytoplasm, and only 8% in the nuclei.
With our current nuclear isolation
method we must admit the possibility
of contamination of cytoplasmic RNA
in our nuclear preparations. Therefore,
we conclude that most, if not all, of the
long-lived mRNA is localized in the
cytoplasmic fraction. The result also
indicates that with our current method
we can eliminate about 90% of un-
labeled fibroin mRNA from the nuclear
preparation. The newly synthesized
fibroin mRNA (in 30 min) was found
about equally in the cytoplasmic frac-
tion (47%) and the nuclear fraction
(51%). This indicates that it takes less
than 30 min to transport fibroin mRNA
from the nucleus to the cytoplasm.
RNA synthesis with disrupted and
isolated nuclei using endogenous RNA
polymerase was carried out at 25 °C.
The incorporation of 3 H-UTP or a- 32 P-
UTO was almost linear for 30 min, and
then slowed, but a slight increase was
still observed at 120 min. The labeled
RNA was usually extracted after 30
min incubation for characterization.
RNA transcribed in vitro was hybri-
dized to B. mori DNA fractionated by
Act D/CsCl (Fig. 20). The hybridization
conditions were such that only trans-
cripts of repetitive sequences, which
include fibroin mRNA because of its
internal repetitive nature, hybridized.
As shown in Figure 20, a prominent
hybridization peak is seen at fractions
9 to 11. This peak is presumed to be
rRNA since it was competed for by cold
rRNA. A small and heterogeneous
peak at fractions 16 to 19, which is the
region for fibroin genes, was competed
for by unlabeled pure fibroin mRNA
(see Table 5). It was also observed that
virtually all of the counts hybridized
were competed for by total posterior
gland RNA. These results indicate that
similar repetitive sequences are trans-
cribed both in vivo and in vitro.
The RNA labeled in vitro for 30 min
was fractionated by a Bio-Gel A-50m
column (Fig. 21a) and compared with
the RNA labeled in vivo for 30 min
TABLE 4. Localization of Fibroin mRNA in Cytoplasmic and Nuclear Fractions
Fractions
Newly Synthesized
mRNA*
(%)
Long-Lived,
Unlabeled mRNAt
(%)
Cytoplasmic sup (a)
Sucrose sup (b)
Nuclear pellet (c)
47
2
51
84
Homogenate of posterior silk glands from the 7th day of the fifth instar was washed and centrifuged
three times in the presence of NP40, and the combined supernates were designated as cytoplasmic sup
(a). Crude pellet thus obtained was centrifuged through 2.2 M sucrose, and the supernate (b) and
nuclear pellet (c) were obtained.
* Posterior silk glands were labeled in vivo for 30 min with :i H-uridine and subjected to the
fractionation mentioned above, and radioactive fibroin mRNA was quantitated by T, RNase
fingerprint method.
tThe long-lived, unlabeled mRNA was quantitated by A 260 oligonucleotide profile after T, RNase
digestion.
DEPARTMENT OF EMBRYOLOGY
41
ro
i
O
u
FRACTION NUMBER
Fig. 20. Hybridization of RNA synthesized in
vitro to B. mori DNA fractionated by Act D/CsCl
and competition of the hybridization by un-
labeled posterior silk gland RNA. The RNA was
labeled with 500 /xCi of a- i2 P-UTP (23.7 Ci/
mmole) in vitro for 30 min in disrupted nuclei
(170 jug of DNA). Aliquots equivalent to 5% of
whole nuclear preparation from a pair of pos-
terior silk glands were hybridized toS. mori DNA
(200 yu.g/gradient) in the presence or absence of
competitor at 50°C for 17 hr in 50% formamide-4
x SET. No competitor added (•); competitor RNA
equivalent to 20-fold excess of posterior-gland
nuclei added (o o); competitor equivalent to
200-fold excess of posterior-gland nuclei added
(A A).
(Fig. 21b). Comparison of Fig. 21a and
Fig. 21b reveals that the radioactivity-
profiles from Bio-Gel columns are simi-
lar for RNA labeled in vitro and in
vivo. However, there is a higher pro-
portion of lower molecular weight
RNAs labeled in vitro than in vivo.
More than 55% of the radioactivity of
the RNA in vitro was found in the re-
gion larger than 28S (Fig. 21a).
The void volume fraction shown in
Fig. 21a, which contained 14% of the
total radioactivity, was recovered and
an aliquot was subjected to RNase T 1
digestion for characterization. The
sample was estimated to contain about
15% fibroin mRNA (Fig. 22c). There-
fore, we conclude that about 2.1% of the
total transcripts synthesized in vitro by
disrupted nuclei in 30 min was fibroin
mRNA of apparent full-size length.
This is the highest value ever found for
a single mRNA species and is about
two orders of magnitude higher than
that of globin mRNA sequences syn-
thesized in vitro. The hybridization
assay was used to confirm the presence
of mRNA sequences. An aliquot of the
void volume fraction of Fig. 21a was
hybridized to DNA in the fibroin gene
region (Fig. 23). The hybrid was recov-
ered from filters 10 to 15, digested with
T j RNase, and the digest was subjected
to a DEAE-Sephadex A25 column (Fig.
22d). Figure 22d demonstrates the
existence of fibroin mRNA sequences
in the RNA in vitro, and a comparison
of Fig. 22c (15% pure fibroin mRNA)
with Fig. 22d (40% pure) indicates that
mRNA was selected by the hybridiza-
tion procedure.
Nucleoplasmic RNA polymerase II is
known to be sensitive to low concentra-
tions of cv-amanitin and is possibly the
TABLE 5. Inhibition of Fibroin mRNA Synthesized in vitro by Low Dose of a-Amanitin
System of RNA Synthesis
Hybridization to Partially Purified Fibroin Genes
Without Cold Fibroin mRNA
With Cold Fibroin mRNA
Complete (a)
Complete + a-amanitin (b)
474 cpm
156 cpm
124 cpm
112 cpm
Nuclei containing 150 fj.g of DNA were incubated with 80 /iiCi of :! H-UTP for 30 min in the absence
(a) or presence (b) of 0.2 /u.g/ml of a-amanitin. For assay 2.0 mg of 5. mori DNA was fractionated by
Act D/CsCl, and 10% aliquots of the fractions containing fibroin genes were used for hybridization.
For the hybridization 2.5% aliquots of each reaction mixture (a or b) were used in the absence or
presence of unlabeled fibroin mRNA (5.1 /i.g/ml). Hybridization was done at 50°C for 2.5 hr in 5 ml of
50% formamide-4 x SET.
42
CARNEGIE INSTITUTION
O
X
o
number
Fig. 21. Bio-Gel fractionation of the RNAs synthesized in vitro by disrupted nuclei (a) and in vivo
(b) in 30 min. (a) On the 7th day of the fifth instar disrupted nuclei were isolated from the posterior
silk glands. An aliquot of the nuclei containing about 200 /xg of DNA was incubated with 250 /^Ci of
a- ,2 P-UTP (10 Ci/mmole) and other ingredients in 1.0 ml volume at 25°C for 30 min. The RNA
products were applied to a 2.6 x 66 cm column of Bio-Gel A-50m, 5.5-ml fractions were collected, and
0.5-ml aliquots were counted after TCA precipitation, (b) A larva was injected with 2.5 mCi of
'H-undine on the 7th day of the fifth instar and incubated at 25°C for 30 min. The RNA was extracted
from the posterior gland and fractionated as in (a). cts/min/0.5ml (• •); A, (i „ monitored by a Gilson
spectrophotometer ( ).
polymerase form that is involved in
transcription of the nuclear precursors
to mRNAs. Therefore, we wondered
whether fibroin mRNA synthesis in
disrupted nuclei is sensitive to low
doses of a-amanitin. RNA synthesis
was carried out for 30 min in the pre-
sence or absence of 0.2 fxg/ml of
a-amanitin. The drug inhibited 21% of
RNA synthesis in vitro. Each product
was hybridized to partially purified
fibroin genes in the presence or ab-
sence of unlabeled pure fibroin mRNA
as a competitor (Table 5). Fibroin
mRNA, as defined by labeled RNA that
hybridized to fibroin DNA and was
competed for by excess cold fibroin
mRNA, was not synthesized in the pre-
sence of a-amanitin, thus providing the
first direct evidence that RNA
polymerase II is the form that trans-
cribes structural genes in the living
cell.
We wanted to test the ability of the
disrupted nuclei to initiate RNA syn-
thesis in vitro. Incorporation of y- 32 P-
ATP was carried out for 30 min. The
RNA was fractionated on a Bio-Gel
A-50m column (Fig. 24). Aliquots of
fractions 14 to 19 were pooled and sub-
jected to the following analysis. The
sample was digested with RNase T t
and the digest was divided into two.
One part was directly fractionated by a
DEAE-Sephadex column (Fig. 25a).
The other was digested further with
alkaline phosphatase and then sub-
jected to a DEAE-Sephadex column
(Fig. 25b). In Fig. 25a, digests having
negative charges of 5.5, 6.5, 7.5 and so
on were observed. These are tenta-
tively identified as 32 ppp A p G p,
32 ppp A p x p G p, 32 ppp A p x p x p G p and so on.
The 32 P-radioactivity shown in Fig. 25a
was mostly sensitive to alkaline phos-
phatase (Fig. 25b). This result excludes
DEPARTMENT OF EMBRYOLOGY
43
e
I H m EY33
a
M
2 4
X
-
'
,
c
E
£ 2
(J
J
y
L
Wvy
-v^.
rf y
d
50
"1
25
1J
ft^V\
*£.
Wf
v \
20
40
Fraction number
Fig. 22. RNase T, "fingerprints" of various RNAs labeled with a-' 2 P-UTP. (a) Pure fibroin mRNA
from the posterior glands labeled in vivo for 24 hr during the 6th day of the fifth instar. (b) High
molecular weight RNA (void volume fraction of a Bio-Gel A-50m column) from animal's carcass
labeled in vivo for 24 hr during the 6th day. (c) An aliquot of crude fibroin mRNA fraction (the void
volume fraction shown by the bracket in Fig. 21) synthesized in vitro in disrupted nuclei, (d) Other
aliquot of the void volume fraction of Fig. 21 was hybridized to fibroin genes (Fig. 23), and purified. 32 P
cts/min (• -•).
the unlikely possibility that a strange
structure like A(5')ppp 32 (5')G p or
A(5')pp P 32(5')X p G p is the product. Frac-
tions 20 to 54 were also pooled, and
analyzed. Hydrolysis of an aliquot of
the sample by 0.3N KOH gave a single
peak at a negative charge of 4.3 (Fig.
25c), which we tentatively identify as
32 ppp A ppp. Therefore, 32 P was not in-
corporated into internal positions. Di-
gests having negative charges of 5, 6, 7,
etc., were obtained by pancreatic
RNase digestion (Fig. 25d) together
with an unidentified peak having -3.8
charge. By T 1 RNase digestion we
found digests having negative charges
of 5.5, 6.5, 7.5, etc. (Fig. 25). There was
a small peak cochromatographing with
5'-ATP marker and a large peak hav-
ing -4.3 charge which we could not
identify (Fig. 25e). Although there are
some unidentified products in the di-
gests, we conclude that most of the
y- 32 P-ATP was incorporated at the 5'
end of RNA molecules in the void col-
umn fraction, which contained 15% fi-
broin mRNA, as well as in the lower
molecular weight region of the Bio-Gel
column (Fig. 24). These observations
that the disrupted nuclei retain their
ability to reinitiate RNA synthesis
strengthen the usefulness of the sys-
tem for the study of fibroin gene regu-
lation in vitro.
44
CARNEGIE INSTITUTION
cvj
i
O
u
5 10 15
FRACTION NUMBER
20
20 40 60
FRACTION NUMBER
Fig. 23. Hybridization of crude fibroin mRNA
synthesized in vitro to partially purified Fibroin
genes. B. mori DNA (2.0 mg) was fully sheared,
actinomycin D (1.0 mg) was added, and the re-
fractive index was adjusted to 1.3842. The mix-
ture was centrifuged at 32,000 rev/min at 20°C
for 66 hr in a Spinco 50.2 rotor, and fractions
were trapped on filters. The void volume fraction
of Fig. 21a was hybridized to the DNA filters in
50% formamide-4 x SET at 50°C for 20 hr, and
the RNA hybridized to fibroin gene region shown
by the bracket was recovered for RNase T,
"fingerprint" analysis (Fig. 22d). In order to ob-
tain enough counts the hybridization and elution
of the hybrid RNA was repeated, cts/min
(• •); A, 60 ( ).
Fig. 24. Incorporation of y- i2 P-ATP into RNA
synthesized in vitro by disrupted nuclei. Nuclei
containing 150 /xg of DNA were incubated with
500 fid of y- :i2 P-ATP (18.7 Ci/mmole) and other
ingredients in 1.0 ml volume at 25 C C for 30 min.
The RNA was phenol-extracted, precipitated
twice with ethanol, passed through a 0.9 x 66 cm
column of Sephadex G25, precipitated again
twice with ethanol, and fractionated by a 2.6 x
66 cm column of Bio-Gel A-50m. Out of 5.5-ml
fractions each 1.0-ml aliquot was counted after
TCA precipitation. Fractions shown by the brac-
kets were pooled and subjected to analyses shown
in Fig. 25.
DEPARTMENT OF EMBRYOLOGY
45
Fraction number
Fig. 25. Characterization of the RNA synthesized in vitro labeled with y- 32 P-ATP. Aliquots of
fractions 14-19 of Fig. 24 were pooled; E. coli tRNA was then added as a carrier, precipitated with
ethanol twice, and the precipitate digested with T, RNase and divided into two. One was applied
directly to a DEAE-Sephadex column (a), and the other was digested further with alkaline phos-
phatase (b). Aliquots of fractions 20-54 of Fig. 24 were pooled, and E. coli tRNA was added and
precipitated with ethanol twice. Aliquots were subjected to digestion with 0.3 A KOH (c), pancreatic
RNase (d), and T, RNase (e). All the digests were fractionated on DEAE-Sephadex A25 columns with
an NaCl gradient 0.10-0.40 M in the presence of 1M urea. Before application to the column the digest
for (c) was neutralized' and carrier oligonucleotides were added. Cold 5'-ATP was added as marker to
the digest for (e). i2 P cts/min (• • ); A, (1 „ ( ); [MJNaCl (- - -).
BIOGENESIS OF MITOCHONDRIA
/. B. Dawid, E. A. Godwin, C. K. Klukas, J. L. Ramirez, S. Ohi, and W. B. Upholt
with the assistance ofM. Rebbert
We have continued our studies on
the structure of mitochondrial DNA
(mtDNA) in several animal cells with
the aim of obtaining comparative maps
of different mtDNAs (see Year Book 74,
p. 46). Ultimately, we would like to
know what functions are coded for by
mtDNA; how these functions are ar-
ranged on the DNA; how the primary
sequence and the overall anatomy of
mtDNA changes in evolution; and
what kind of evolutionary forces and
mechanisms operate on this DNA.
MtDNA codes for two specific
ribosomal RNAs (rRNAs), a set of
transfer RNAs (tRNAs), and several
poly(A)-containing RNAs that may be
messenger RNAs. The rRNAs and
46
CARNEGIE INSTITUTION
tRNAs have been well characterized in
a number of organisms. Our aim has
been to map the location of the regions
coding for these RNAs in the mtDNA of
Xenopus laevis and X. mulleri. We
have also studied the location of the
D-loop, which marks the replication
origin. Experiments have been ini-
tiated to study nonribosomal RNAs in
Xenopus mitochondria. A mapping
study of Drosophila melanogaster
mtDNA has been completed during
this year. A detailed comparison of the
mtDNAs of the goat and sheep has
been carried out.
The major results of these experi-
ments are as follows. The mt-rRNA
genes in Xenopus and Drosophila as
well as in Hela mtDNA are located
next to each other on the H-strand with
a gap of 120 to 160 nucleotides between
them. In X. laevis and Hela this gap
carries the gene sequence for a
mitochondrial 4S RNA, and one 4S
RNA gene is located immediately be-
yond the ends of the rRNA genes. A
total of 15 to 16 genes for 4S RNA are
distributed widely over the H-strand.
These features are again similar to the
situation in Hela mtDNA as reported
by Wu and her colleagues {Journal of
Molecular Biology, 71, 81-93, 1972).
We further determined the polarity of
the mt-rRNA in Drosophila and
Xenopus: transcription proceeds from
the smaller to the larger RNA. Trans-
cription in the direction small-to-large
rRNA is now recognized to be generally
true for rRNA genes. In both X. laevis
and X. mulleri the D-loop begins at a
distance of about 800 base pairs from
the end of the rRNA genes and replica-
tion is directly away from the rRNA
genes.
We reported previously {Year Book
71, p. 22; Year Book 70, p.44) that the
primary sequence of mtDNA changes
quite rapidly in evolution. Our recent
observations show that the overall
anatomy of the mtDNA molecule is
highly conserved over long evolu-
tionary distances.
Cloning OFXenopus laevis
Mitochondrial DNA
/. B. Dawid, W. B. Upholt, and J. L. Ramirez
with the assistance ofM. Rebbert
To aid structural studies of X. laevis
mtDNA, we prepared hybrid plasmids
between mtDNA and suitable vectors
and grew these plasmids in E. coli (see
Year Book 74, pp. 16 and 20). In this
work we adhered to the safety stan-
dards proposed by the Asilomar Con-
ference on Recombinant DNA. The re-
striction endonuclease EcoRI cleaves
X. laevis mtDNA into two fragments,
one of 15 kb and one of 2 kb (kb is
kilobases or kilobase pairs). The small
fragment was linked to the
kanamycin-resistant vector PCR11
(PCR11 is 12.6 kb); the larger fragment
was linked to the tetracycline-resistant
vector PMB9 (PMB9 is 5.3 kb). After
growth in E. coli the recombinant
plasmids were isolated, digested with
EcoRI, and the mitochondrial frag-
ments were purified. Both cloned
fragments were shown to be identical
with the EcoRI fragments of frog
mtDNA by gel electrophoretic analysis
of the patterns obtained by digestion
with other restriction enzymes. These
cloned mtDNA fragments were then
used in mapping studies in addition to
fragments isolated directly from frog
mtDNA.
Maps of X. laevis and X. mulleri
mtDNA: Restriction Sites, the
D-loop, and the rRNA Genes
J. L. Ramirez and I. B. Dawid
In Year Book 74, pp. 46-47, we
showed maps for these two mtDNAs
which contain the locations of EcoRI
sites, the Smal site in X. mulleri
mtDNA, and the D-loop positions. We
now add the positions and the polarity
of the rRNA genes (Fig. 26).
DEPARTMENT OF EMBRYOLOGY
47
Bam \
Fig. 26. Maps ofX. laeuis andX. mulleri mtDNA. EcoRI sites are numbered in italics inside the
circles. Other restriction sites are labeled. The rRNA genes and the D-loops are shown, and the
polarity of transcription and replication is indicated with an arrow pointing in the 5'-to-3' direction
on the growing RNA or DNA strand. Distances are given as percent of total genome, which is 17,100
base pairs. The H-strand is shown on the outside, the L-strand on the inside.
The mapping of the rRNA genes was
done in the following way. We know
that the two genes are located next to
each other with a gap of about 120
bases separating them {Year Book 72,
p. 45). EcoRI fragments of X. laevis
mtDNA were annealed with mt-rRNA,
and the hybrid molecules were in-
spected in the electron microscope. The
rRNA/DNA hybrid regions were close
to the end of the larger EcoRI frag-
ment, with the small mt-rRNA located
at the end. This position for the rRNAs
was confirmed by "R-loop mapping," a
method developed by White and Hog-
ness of Stanford University. In this
method, native DNA and RNA are an-
nealed under conditions in which the
RNA invades the DNA duplex and dis-
places one of the strands. A loop is
clearly visible in the electron micro-
scope. To determine the relation of the
rRNA genes and the D-loop, the DNA
is fixed with glyoxal to preserve the
D-loop, then digested with EcoRI, and
finally hybridized with large rRNA be-
fore examination in the electron micro-
scope. Figure 27 shows such a molecule
which established the relative posi-
tions of the D-loop and the rRNA genes
in X. laevis mtDNA. The X. mulleri
map was derived in an analogous way
and is also shown in Fig. 26.
The polarity of the mt-rRNA was de-
termined by digestion of EcoRI frag-
ments of the DNA with specific exonu-
cleases, followed by hybridization of the
resulting single-stranded tails with
rRNA. The result is incorporated into
the models of Fig. 26. The direction of
replication was deduced from the ob-
servation of a number of "expanded
D-loop molecules" in X. mulleri
mtDNA, and is also shown in the maps.
48
CARNEGIE INSTITUTION
■ v "'>^ ^
Fig. 27. Electron micrograph and tracing of an X. laeuis mtDNA molecule with the D-loop in the
"H-form" ID) and an R-loop formed by the large rRNA (R). Note that the R-loop has a "whisker," i.e., a
portion of the RNA is displaced by the DNA strand and forms a single-stranded tail.
Mapping of 4S RNA Genes on
X. laeuis mtDNA
S. Ohi, W. B. Upholt and I. B. Dawid
In Year Book 74, p. 47, we described
our initial experiments on the mapping
of mitochondrial 4S RNA genes of X.
laeuis. The covalent attachment of fer-
ritin to 4S RNA was achieved by the
method of Wu and Davidson (Journal
of Molecular Biology 78, 1, 1973);
ferritin-4S RNA complexes were hy-
bridized to separated strands of
mtDNA, and the hybrids were
analyzed in the electron microscope.
The H-strand of mtDNA carries the
genes for the mt-rRNAs and a majority
of the 4S RNA genes. The rRNA re-
gions were used as reference points in
the mapping. The result of the meas-
urements of these hybrid molecules is
shown in Fig. 28. At least 15 and possi-
bly 16 sites for 4S RNA were found on
the H-strand. The more remarkable
positions are those between and im-
mediately flanking the rRNA genes:
Hela cell mtDNA has 4S RNA posi-
tions at the analogous points, as de-
termined by Wu and her colleagues (J.
Mol. Biol. 71, 71-93,1972). The remain-
ing 4S RNA sites are distributed
widely over the H-strand. This wide
distribution is also similar to the situa-
tion in Hela mtDNA.
Our results with the L-strand of X.
laeuis mtDNA are incomplete. There is,
however, tentative evidence for 5 or 6
sites on this strand. If these results can
be confirmed in future experiments, we
would have evidence for 20 to 22 sites
for 4S RNA on Xenopus mtDNA. This
number would come close to the ex-
pected number of tRNAs required for a
complete protein synthesizing "ma-
chine" that depends fully on endogen-
ous tRNAs.
DEPARTMENT OF EMBRYOLOGY
49
Fig. 28. Map of 4S RNA sites on the H-strand of X. laevis mtDNA. Closely adjacent sites are only
then considered to represent two sites if both sites were seen repeatedly to be occupied by ferritin on
the same DNA strand. Site 10 may be a doublet site, but it has been observed only twice out of 158
molecules examined as a doublet and is not considered certain. Comparison with Fig. 26 shows that the
D-loop position is from about 0.82 to 0.90 on this map. Thus, position H14 may be within the D-loop,
but this is not certain.
Mapping Studies with mtDNA
from Drosophila melanogaster
C. K. Klukas and I. B. Dawid
mtDNA and mt-rRNA from
Drosophila have been isolated and
characterized {Year Book 74, pp. 49-
53). The circular mtDNA molecule has
a molecular weight of 12 x 10 6 , i.e., it
consists of 18,400 base pairs. About
25% of this molecule has a very high
content of adenylic and thymidylic
acids and can be visualized as an early
melting region by electron microscopy.
The mt-rRNAs have molecular weights
of about 300,000 and 530,000.
We have mapped the positions of the
cleavage sites of the restriction en-
donucleases Haelll and Hindlll rela-
tive to each other and relative to the
early melting region. The map was
constructed from electron microscopic
measurements of partially denatured
total and partial restriction digestion
of the DNA, and from gel electrophore-
tic analysis of the DNA with each en-
zyme separately and both enzymes to-
gether. Figure 29 shows the results.
We next mapped the location of the
rRNA genes relative to restriction sites
by electron microscopy of RNA/DNA
hybrids. The assignment of positions to
the rRNA genes was confirmed and re-
fined by a quantitative hybridization
experiment in which the small and the
large rRNA were separately hybridized
to Hindlll and Haelll fragments of
mtDNA. The resulting map shows the
two rRNA genes located in close prox-
imity and separated by a gap of 160
nucleotides.
The polarity of the rRNA was deter-
mined in an analogous way to the
method outlined forX. laevis mt-rRNA.
As in the case of Xenopus the direction
of transcription in Drosophila mtDNA
is from the smaller to the larger rRNA.
50
Both the close proximity of the two
rRNA genes and the polarity of the
RNA are highly conserved in the evo-
lution of animal mtDNAs, since the
identical arrangement is found in frogs
and fruit flies.
280
Fig. 29. Map of D. melanogaster mtDNA. The
cleavage sites of Haelll and Hindlll are indi-
cated by arrows, and the early melting region is
indicated by an open section of the circle. The
rRNA genes are indicated and are separated by a
gap of 160 nucleotides. The direction of trans-
cription, i.e., 5' to 3' polarity of the RNA, is
shown. Distances are given in base pairs. The
total genome size is 18,400 base pairs.
Mapping and Sequence
Comparisons of Goat and
Sheep mt DNAs
W.B. Upholt
A detailed comparison of sheep and
goat mtDNA was initiated with the
aim of observing the differences in the
mtDNAs of closely related animals
(Year Book 74, p.48). This study should
give information about the least con-
served regions of the mtDNA rather
than the most conserved ones, which
CARNEGIE INSTITUTION
are best studied in more distantly re-
lated species (see Year Book 70, p. 44).
The analysis of restriction fragments
of sheep and goat mtDNA was carried
further. Endonuclease Haelll yields 27
to 30 fragments from these mtDNAs.
From a consideration of conserved
fragments it follows that sheep and
goat mtDNAs have diverged to the ex-
tent of 7% to 11% of their sequences. In
the course of these studies the interest-
ing fact was observed that the mtDNAs
from individual sheep and goats dif-
fered from each other with respect to
the Haelll fragment pattern and there-
fore with respect to primary sequence.
Figure 30 shows the electrophoretic
separation of these fragments from the
mtDNA of two goats and three sheep.
No two patterns are exactly alike. The
sheep patterns are all rather similar to
each other, the two goats differ sub-
stantially, and the sheep and goat pat-
terns are very different. On the basis of
these experiments and with certain as-
sumptions, one may calculate a se-
quence divergence of 0.5% to 1% among
the sheep mtDNAs and about 2% be-
tween the DNAs of the two goats.
Detailed maps were constructed for
the cleavage sites of endonucleases
EcoRI and Hindlll on both sheep and
goat mtDNAs. The two goats differed
with respect to EcoRI and Hindlll sites,
a fact that emphasizes the considerable
difference in sequences of the two
DNAs. The position and lengths of the
goat and sheep D-loops were deter-
mined and are also shown on the maps
of Fig. 31. Several EcoRI Hindlll sites
were conserved between sheep and
goat, but some of the sites differ. The
D-loops are at analogous positions on
the two mtDNAs, but the lengths of the
loops are different. We reported last
year that a region of high evolutionary
change overlaps one end of the D-loop
(the left end in Fig. 31). This region
where the two D-loops differ in length
and the area immediately beyond it
have been studied in more detail by
electron microscopic heteroduplex
mapping. We confirm the earlier con-
clusion that this region shows a high
DEPARTMENT OF EMBRYOLOGY
51
— G—
P
t- ■ » i
I
^ „
•*>,-■.->,
Fig. 30. Haelll fragments of sheep and goat mtDNA. The direction of electrophoresis is from top to
bottom. Fragment sizes range from just under 2000 to 200 base pairs. Smaller fragments occur but are
not shown in this figure. mtDNA fragments from three individual sheep (S) and two goats (G) are
displayed; the unmarked lane contains Haelll fragments of the plasmid PSC101. Differences in the
band patterns of individual sheep and goats are apparent.
rate of evolutionary change and
further suggest that it contains a dup-
lication of about 150 base pairs. Thus,
the segment of DNA overlapping and
adjacent to the D-loop may show a high
rate of base mutation and also a ten-
dency to rearrangements on a larger
scale.
52
CARNEGIE INSTITUTION
Fig. 31. Maps for sheep (S) and goat (G) mtDNAs. EcoRI sites are labeled/? and numbered; Hindlll
sites are labeled//; fragments are labeled with capital letters. Subscripts refer to sites or fragments
that are present in one of the two goat mtDNAs only. The position and length of each D-loop are
indicated. The lengths of the different regions in the molecules are shown on the outer circles in base
pairs.
PHOSPHOLIPID-CHOLESTEROL DYNAMICS IN MAMMALIAN
CELLS
R. E. Pagano, L. Huang, K. Ozato and M. Takeichi
with the technical assistance ofE. Asch, E. Someruille and D. Duncan
Work in this laboratory is concerned
with the possible dynamic role that
membrane phospholipids and choles-
terol may play in the regulation of cell
behavior. During the past several
years we have been developing a tech-
nique utilizing artificially generated
lipid vesicles, which permits a con-
trolled and systematic perturbation of
these membrane components in living
cells. In the present report we empha-
size our recent findings on the interac-
tions of phosphatidyl choline vesicles
with murine lymphocytes and the
physiological consequences of these in-
teractions. We show that the pathways
of vesicle-cell interactions, enumerated
in previous reports (vesicle-cell fusion,
adsorption, and lipid exchange) are ex-
quisitely sensitive to the molecular
composition and physical properties of
the artificial vesicles. When lympho-
cytes are the "recipient" cells, the re-
sulting membrane perturbations lead
to altered surface properties and an
enhanced proliferative activity of these
cells, when stimulated by various
mitogens and antigens.
In a separate but related line of re-
search, Takeichi has explored some of
the adhesive properties of a fibroblastic
cell line and reports on attempts to
identify and characterize membrane
proteins that participate in this
phenomenon. Parallel studies on the
adherence of lipid vesicles of known
composition to cells are also discussed.
Finally, a critique of the use of small
DEPARTMENT OF EMBRYOLOGY
53
hydrophobic fluorescent probes to
monitor changes in the physical state
of the lipids in natural membranes is
presented.
Details of our studies on each of
these topics are presented in separate
sections below.
Interactions of Phospholipid
Vesicles with Murine
Lymphocytes: Vesicle-Cell
Adsorption and Fusion as
Alternative Pathways of
Uptake
L. Huang, K. Ozato, and R. E. Pagano
The various mechanisms by which
unilamellar lipid vesicles interact with
cells and the subsequent modification
of cell behavior that would result have
been the subject of recent investiga-
tions from this and other laboratories.
In Year Book 74 (p. 53) we showed that
the incorporation of lecithin into
Chinese hamster fibroblasts from
dioleyl lecithin (DOL) vesicles were
primarily via vesicle to cell fusion and
lipid exchange processes. The interac-
tion of dimyristoyl lecithin (DML) vesi-
cles with these cells was also studied,
and exhibited an enhanced uptake
when the temperature of vesicle-cell
incubation was below the thermotropic
phase transition temperature (T c —
25 °C) of DML. While the detailed
mechanisms for DML uptake were not
explored, this peculiar temperature
dependence suggested an additional
pathway for vesicle incorporation. In
the present study with mouse thymic
lymphocytes, we identify this pathway
as vesicle-cell adsorption. Further-
more, in a survey of the uptake of vari-
ous lecithin vesicles by thymocytes, it
is shown that these lipid vesicles can be
classified according to whether they
are incorporated principally by vesicle
to cell fusion or a stable vesicle to cell
adsorption.
Evidence for a Stable Vesicle-Cell
Adsorption
In an EM autoradiographic survey of
thymocytes treated with a variety of
[ 3 H] lecithin vesicles, we observed two
distinct patterns of silver grain distri-
butions in treated cells. In the case of
EYL (egg yolk lecithin) or DOL vesi-
cles, most of the exogenous lipid label
appeared inside the cell, while for DML
or DPL (dipalmitoyl lecithin) vesicle-
treated cells, relatively large amounts
of radioactivity appeared to reside at
the cell surface. Typical examples of
such autoradiograms are shown in Fig.
32 for cells treated with EYL vesicles
at
3
Kf> &
a
A*
^
%j£
•4*.™
4
&
J
«>'
S*
*
Fig. 32. Electron microscope autoradiographs of mouse thymocytes treated with (a) [ 3 H] DML
vesicles (2°C, 60 min), x 14,000 and (b) [ :i Hj EYL vesicles (22-23°C, 30 min), x 7,700. Bar is 0.1 /xm.
54
CARNEGIE INSTITUTION
(30 min at 22-23 °C) and DML vesicles
(1 hr at 2 °C). Similar auto-radiograms
were also obtained for thymocytes
treated with these lipids at 37 °C. Thin
layer chromatographic (TLC) analysis
of the extracted lipids from these cells
revealed that most (—90%) of the
radiolabel was still in lecithin, indicat-
ing that the grains seen in the au-
toradiograms represent the approxi-
mate location of the exogenously intro-
duced lecithin and not some degrada-
tion product. This qualitative observa-
tion of two classes of grain distribution
was confirmed in a statistical analysis
of a large number of autoradiograms of
treated cells and is summarized in
Table 6. The accumulation of silver
grains at the surfaces of DPL or DML
(vs. EYL or DOL) vesicle-treated cells
is readily apparent, both from calcula-
tions of the percentage of grains found
near the cell surface and from the ratio
of the normalized grain densities.
It should be noted that all vesicle-cell
incubations were carried out at 22-
23 °C except in the case of DML. In this
instance, incubations were carried out
at 2 °C to avoid complications which
might arise from the phase transition
of the DML vesicles. A suggestion of
the complex uptake behavior which re-
sults from this transition is seen in Fig.
33, where the temperature dependence
of DML and EYL uptake by thymo-
cytes is compared. Each data point in
this figure represents the lipid uptake
by cells following a 60-min incubation
at a given temperature. For EYL vesi-
cles, the lipid uptake increased mono-
tonically with temperature between
0°C and 37°C. In striking contrast, the
lipid uptake from DML vesicles showed
a peculiar temperature dependence,
exhibiting a sharp maximum at about
20°C and minima at about 15° and
24°C. Furthermore, the uptake at 0°C
was greater than at 37°C. DML uptake
was always larger than EYL uptake;
for example, about 1.5 ^ig of EYL vs. 4
/u,g of DML were taken up by 10 7 cells
at 37°C.
10 20 30 40
Temperature (°C)
Fig. 33. Temperature dependence of J :i H] EYL
(1 mg/ml, 1 x 10" cpm/mg) and [ 14 Cj DML (1
mg/ml, 4 x 10 s cpm/mg) uptake by mouse thymo-
cytes. 10 7 cells/ml were incubated 60 min with
vesicles in Gey's balanced salt solution.
The existence of two types of lecithin
vesicle-cell interactions suggested by
the data above was confirmed in ul-
TABLE 6. Distribution of Autoradiographic Grains over Mouse Thymocytes
Treated with [ :! H] Lecithin Vesicles
(Surface Grain*
Treatment
Percent Grains
in Surface
Density)
. + SEM
Time
Temp.
Total Silver
(Interior Grain
Lipid Type
(min)
CO
Grains Counted
Compartment*
Density)
:! H DML
60
2
565
69
8.1 ± 1.0
<H DPL
30
22-23
342
71
8.6 ± 1.2
'H EYL
30
22-23
348
27
1.5 ± 0.2
:! H DOL
30
22-23
133
25
1.4 ± 0.2
'See Pagano, Ozato, and Ruysschaert, elsewhere in this report, for description of analysis.
DEPARTMENT OF EMBRYOLOGY
55
trastructural studies of vesicle-treated
cells. Thymocytes incubated with EYL
or DOL vesicles at temperatures be-
tween 2 ° and 37 °C showed smooth cell
surfaces identical to those of untreated
control cells. A typical example of the
surface of an EYL-treated cell is shown
in Fig. 34a. In the DPL-treated cells
(Fig. 34b), many of the intact vesicles
had apparently adsorbed to the cell
surface. These vesicles ranged in size
from about 250 A to about 650 A in
diameter, possibly reflecting heteroge-
neity in the applied vesicle population.
Furthermore, the distribution of ad-
sorbed vesicles on the cell surface was
not uniform. Micrographs obtained
from cells treated with DML vesicles
also showed evidence of adsorbed ma-
terial at their surfaces. However, the
DML vesicles were apparently rup-
tured or collapsed in the 2 °C treat-
ments (Fig. 34c), while large masses of
amorphous material were seen ad-
sorbed to cells treated at 37 °C (Fig.
34d). From the ultrastructural studies
shown in Fig. 34 and the autoradio-
graphic data (Fig. 32; Table 6) showing
an accumulation of [ [6 H] DPL or [ 3 H]
DML at the surfaces of treated cells, it
is concluded that the adsorbed mate-
rials seen in Fig. 34b, c, d actually rep-
resent native or modified applied vesi-
cles. In the case of DML, intact vesicles
were rarely seen adsorbed at the sur-
face of treated cells. The presence of
collapsed or ruptured vesicles and
amorphous material suggests a reor-
ganization of the vesicle membranes.
This modification of vesicle morphol-
ogy could be a consequence of an inher-
ent bilayer instability in this relatively
short chain lecithin. Our findings of an
enhanced uptake of DML by thymo-
cytes near or below the phase transi-
tion temperature (T c ) or DML (Fig. 33)
and the ultrastructural evidence pre-
sented in Fig. 34c and d suggest an en-
Fig. 34. Transmission electron micrographs of mouse thymocytes treated for 30 min with vesicles
comprised of (a) EYL (22-23°C), (b) DPL (22-23°C), (c) DML (2°C) and (d) DML (37°C) x 78,000. Bar is
1 pirn. Intact vesicles (b), collapsed or ruptured vesicles (c), and amorphous material (d) are seen at
arrows.
56
CARNEGIE INSTITUTION
hanced adsorption of DML at these
temperatures. Such an enhancement
could be due to the increase in bilayer
thickness below the T c of the lipid.
Vesicle-Cell Fusion and Lipid
Exchange
Since the EM autoradiographic data
as well as the micrographs presented in
Fig. 34a suggested an absence of
vesicle-cell adsorption for DOL and
EYL vesicles, other possible mecha-
nisms for the uptake of these vesicles
have been investigated. In Year Book
74 (p. 62) we suggested that EYL vesi-
cles were incorporated by thymocytes
principally by a vesicle-cell fusion
mechanism. This suggestion was based
in part on the finding that when
thymocytes were treated with [ 14 C]
EYL vesicles containing trapped [ 3 H]
inulin, both components became cell
associated at identical rates, consistent
with the uptake of "intact" vesicles.
Additionally, the uptake of EYL vesi-
cles was insensitive to NaN 3 and
cytochalasin-B, two known inhibitors
of endocytosis in other cell types. Dur-
ing the past year, further direct evi-
dence for the fusion of EYL vesicles
with thymocytes was obtained. Cells
were treated with EYL vesicles con-
taining high specific activity [ 3 H] in-
ulin under various conditions. EM au-
toradiography was then used to
examine the intracellular distribution
of the vesicle marker in treated cells.
Statistical analysis of a large number
of autoradiograms showed approxi-
mately 75% of the trapped PH] inulin
was randomly distributed in the cyto-
plasm and nucleoplasm of the treated
cells, suggesting that the contents of
EYL vesicles were released into the
cells by fusion.
In order to explore the possibility of
lipid exchange between vesicles and
cells, thymocytes with endogenously
( 3 H-palmitate) labeled lipids were in-
cubated with [ 14 C] EYL unilamellar
vesicles for 1 hr at 37 °C. Following
this incubation, the vesicles were sepa-
rated from the cells and subjected to
molecular sieve chromatography on
Sepharose 4B. Each fraction was
examined for both 14 C- and 3 H-cpm.
The results of such an experiment are
given in Fig. 35. The 14 C-cpm showed
one broad symmetrical peak, typical of
unilamellar lipid vesicles. Three
3 H-cpm peaks were observed. The first
peak eluted at the exclusion volume of
the column and probably represents
20 30 40
FRACTION NUMBER
Fig. .'35. Elution profile of ["C] EYL vesicles on Sepharose 4B following incubation (1 hr at 37°C)
with mouse thymocytes. Endogenous cell lipids were labeled with [ :i Hj palmitic acid. Fractions
between arrows were analyzed for lipid composition.
DEPARTMENT OF EMBRYOLOGY
57
unremoved whole cells or fragments.
The third and major 3 H-cpm peak
eluted at the end of the column volume;
its radioactivity was chloroform-
methanol nonextractable and presum-
ably represents some secreted
metabolic products of small molecular
weights. The second 3 H-cpm peak pre-
cisely cochromatographed with the un-
ilamellar 14 C-vesicle peak. Between
fractions 15 and 27 (Fig. 35, arrows) a
total of 3 x 10 3 3 H-cpm was found. In
control experiments in which the [ 14 C]
EYL vesicles were incubated with a
balanced salt solution previously in
contact with the 3 H labeled cells, a
total of only 5 x 10 2 3 H-cpm was found
in these fractions. Practically all
(-90%) of both the 3 H-cpm and 14 C-cpm
in these fractions was chloroform-
methanol extractable; TLC analysis
shows more than 80% of the 3 H-cpm
was lecithin. These results demon-
strate that transfer of cellular lipid to
the vesicles, always present in excess,
can occur in addition to EYL vesicle-
cell fusion. In this experiment, al-
though lecithin was a minor lipid
(—30%) in the endogenously labeled
thymocytes, it constituted a major frac-
tion (—80%) of the cell lipids trans-
ferred to the vesicles. This indicates an
apparent specificity in the vesicle-cell
lipid transfer process. From the data in
Fig. 35, the number of lecithin
molecules per EYL vesicle, and a lower
limit for the specific activity of the cel-
lular lecithin, it was calculated that no
more than two cellular lecithin
molecules were transferred per 100
EYL vesicles. If it is assumed that the
transfer of cell lipids to vesicles is ac-
companied by an equal amount of
transfer of vesicle lipids to the cells, it
is further calculated that only 1-2% of
the total EYL uptake (37°C) is due to
the lipid exchange process.
Thus we conclude that the major
pathways of lecithin vesicle uptake by
mouse thymocytes are (1) a stable vesi-
cle to cell adsorption in the cases of
DML or DPL and (2) vesicle-to-cell fu-
sion for EYL (or DOL) with a minor
contribution due to lipid exchange.
Since all these processes require an ini-
tial vesicle-to-cell contact, it follows
that for DML and DPL this contact is
further developed into a state of stable
adsorption, while for EYL or DOL it
leads to subsequent fusion. We specu-
late that the physical properties of the
lipid vesicles, dictated by the fatty acyl
composition of the vesicle lecithin, ul-
timately determine the predominant
pathway for vesicle uptake.
Interaction of Phospholipid
Vesicles with Murine
Lymphocytes: Vesicle-Cell
Adsorption Leads to Altered
Surface Properties and
Enhanced Proliferative
Response
K. Ozato, L. Huang, and R. E. Pagano
In order to delineate the relationship
between cell surface properties and
membrane lipid composition, we have
explored the use of unilamellar lipid
vesicles to perturb the surface of mouse
lymphocytes, stimulated by various
mitogens to undergo blast transforma-
tion. This approach seemed particu-
larly hopeful, since previous studies
from this and other laboratories using
other cell types had demonstrated that
vesicle lipids could be incorporated into
cells by vesicle-to-cell fusion and lipid
exchange processes. We first made a
detailed study of the interactions of a
variety of lecithin vesicle types with
mouse lymphocytes and confirmed the
existence of these pathways of vesicle
uptake (Huang, Ozato, and Pagano;
this report). Surprisingly, in certain
types of lipid vesicles, a previously un-
reported mechanism of vesicle uptake
was also found, namely, a stable
vesicle-to-cell adsorption. We report
here on the ability of lipid vesicles to
modify the agglutinability and lectin
binding of lymphocytes. Most signifi-
cantly it is shown that vesicle pre-
treatment can also enhance .the pro-
liferative activity of stimulated cells.
The extent of these modifications of cell
58
CARNEGIE INSTITUTION
behavior depends on the lipid vesicle
type and is correlated directly with the
ability of vesicles to adsorb to the cell
surface.
Enhancement of Proliferation by Lipid
Vesicle Pretreatment
Cortisone-resistant thymocytes
(CRT) were briefly (30 min) treated
with unilamellar lipid (lecithins or 1:2
lecithin: cholesterol) vesicles and sub-
sequently cultured in a serum- and
lipid-free medium containing a
mitogenic dose of Con A (1 fxg/ml). The
proliferative activity of these cultures
was periodically examined using the
criteria of increased PH] TdR incorpo-
ration and increased number of total
cells and lymphoblasts. Figure 36
shows the time course of DNA synthe-
sis in CRT, activated by Con A, for both
DML vesicle pretreated and, for con-
trol, untreated cells. In both cultures
[ 3 H ] TdR incorporation reached a peak
on day 3. The pretreated cells showed
obviously higher DNA synthesis than
control at every time point tested, with
the largest differences (2-3 x ) occurring
at days 3 and 4 in culture. By the fifth
day, DNA synthesis in both cultures
had practically ceased. The degree of
enhancement of the mitogenic response
by DML varies from experiment to ex-
periment (2-5 x control) but was always
the largest of the lipid vesicles tested.
The level of the DML effect was depen-
dent on the temperature of the vesicle
pretreatment with pretreatment at
2 °C giving about a twofold larger ef-
fect than vesicle-cell incubations car-
ried out at 22-23 °C. Table 7 compares
the effect of different lipid vesicle pre-
treatments on the level of DNA syn-
thesis in Con A-stimulated CRT. Each
of the vesicle pretreatments also re-
sulted in an increase in total cells as
well as lymphoblasts in the culture.
The degree of enhanced proliferation
depended on the type of lipid vesicle
used, with DML always giving the
largest stimulation and EYL the least.
▲ DML PRETREATMENT
• CONTROL
12 3 4 5 6
DAYS IN CULTURE
Fig. 36. Time course of DNA synthesis in cortisone-resistant thymocytes stimulated by Con A (1
/xg/ml).
DEPARTMENT OF EMBRYOLOGY
59
TABLE 7. Effects of Lipid- Vesicle Pretre'atments on Cortisone-Resistant
Thymocytes Stimulated with Con A
Temperature of
Typical Experiment
Total/
Lymphoblast/
Lipid Vesicle
Pretreatment
(cpm/culture
Culture
Culture
Pretreatment*
(0°)
x 10- ;! t
x 10""
x 10-
EYL
22-23
11.5 ± 0.9
3.0 ± 0.1
6.1 ± 0.2
DOL
22-23
11.4 ± 0.8
3.0 ± 0.2
6.0 ± 0.3
1:1 EYL:Chol
22-23
14.6 ± 1.2
3.1 ± 0.2
6.3 ± 0.3
DPL
22-23
29.0 ± 1.5
3.7 ± 0.3
9.0 ± 0.5
DML
22-23
18.3 ± 1.1
3.6 ± 0.3
8.7 ± 0.4
DML
2
37.8 ± 1.3
4.4 ± 0.3
14.3 ± 0.8
Control: no vesicles
10.5 ± 0.4
2.8 ± 0.1
5.2 ± 0.3
Control: no mitogen
0.4 ± 0.0
1.7 ± 0.0
0.0
no vesicles
Control: DML or DPL
22-23
0.9 ± 0.0
1.6 ± 0.0
0.0
vesicles;
no mitogen
*Optimum lipid concentration of vesicle suspension was 0.5 mg total lipid/ml cell suspension, except
for 1:1 EYL:Chol vesicles (1.0 mg/ml).
tMean of triplicate cultures ± standard deviation.
The observed order was DML (2-5 x ) >
DPL (2-3 x )> 1: 1 EYL:Chol ( 1.5-2.5 x )
> EYL = DOL (1.2-2 x) > untreated
cells.
Cells pretreated with lipid vesicles
and subsequently cultured in the ab-
sence of mitogen did not exhibit signif-
icant DNA synthesis 72 hr after cul-
ture. Additional experiments in which
cells were continuously cultured in the
presence of lipid vesicles at sufficiently
low concentrations (0.1 mg/ml) to pre-
vent toxicity even for periods of incuba-
tion up to 72 hr, also showed no signifi-
cant enhancement of DNA synthesis
levels. Thus it is concluded that none of
the lipid vesicles tested were them-
selves mitogenic for T-cells.
The effects of lipid vesicle pretreat-
ment on spleen cell cultures stimulated
by a B-cell mitogen, bacterial
lipopolysaccharide (LPS), were also
studied. In nearly every case lipid-
vesicle pretreatment of B-cells resulted
in a significant enhancement of their
proliferative activity when stimulated
by LPS. Again, a graded enhancement
in DNA synthesis, depending on the
vesicle lipids used, was observed in the
same order as that found in T-cells.
Lipid vesicle pretreatment of spleen
cell cultures was not itself mitogenic.
Alterations in Lectin-Induced
Agglutinability and Lectin-Binding in
Vesicle-Treated Cells
The lectin-induced agglutinability of
normal thymocytes pretreated with
lipid vesicles was assayed quantita-
tively by measurements of the total
particle number in cell suspensions in-
cubated with and without Con A. As
shown in Table 8 pretreatment of
thymocytes with lipid vesicles prior to
incubation with Con A significantly
enhanced the agglutinability of the
treated cells. Furthermore, the degree
of enhanced agglutinability was de-
pendent on the vesicle type and showed
exactly the same order as that found in
the proliferation studies, that is, DML
> DPL > 1:1 EYL:Chol > EYL ~ DOL
> no treatment. Vesicle pretreatments
themselves did not induce thymocyte
agglutination in the absence of lectin.
In control and EYL or 1:1 EYL:Chol
vesicle pretreated cells, a-MM (but not
galactose) could almost completely re-
verse Con A-induced agglutination.
However, the agglutination was only
partially reversed for DML or DPL pre-
treatments. These effects of vesicle pre-
treatments on lectin induced aggluti-
nation were consistently observed for
60
CARNEGIE INSTITUTION
different lectin concentrations (5-50
ixg/m\). The effect of vesicle pretreat-
ments on the binding of radiolabeled
Con A to normal thymocytes was also
studied. In Fig. 37, the binding of [ 125 I]
Con A (1 /Ltg/ml; 60 min at 2 °C) to nor-
mal thymocytes pretreated with vari-
ous lipid vesicles is compared. The
binding of [ 125 I] Con A to 1:1
EYL:Chol, EYL or DOL vesicle pre-
treated cells was indistinguishable
from that of untreated control cells,
both in terms of the total as well as the
«-MM inhibitable portion of the bind-
ing. In the case of DML or DPL vesicle
pretreatments, however, the binding
was significantly enhanced. Cells pre-
treated with DPL vesicles always
showed greater total [ 125 I] Con A bind-
ing than DML-pretreated cells, while
the opposite order was found when the
a-MM inhibitable portions were com-
pared. Binding studies carried out at
37 °C showed increased lectin binding
and an identical trend with respect to
the ability of different vesicle types to
modify that binding. The binding of
PHI Con A to vesicle-treated cells
studied at higher lectin concentrations
(50 /xg/ml) also showed a striking en-
hancement of overall and a-MM in-
hibitable binding of DML and DPL
vesicle-pretreated cells.
TABLE 8. Effect of Lipid- Vesicle Pretreatment
on Lectin-Induced Agglutinability of
Thymocytes
Agglutination Index*
Con A (10 pig/ml)
Vesicle Pretreatment
-MM
-MM
None
DML(2°C)
DML(22-23°C)
DPL(22-23°C)
1:1 EYL:Chol
(22-23°C)
DOL(22-23°C)
EYL(22-23°C)
76.5 ± 1.7t 1.0 ± 0.0
93.0 ± 2.4 25.8 ± 0.2
91.8 ± 1.5 12.3 ± 0.1
89.7 ± 1.8 31.6 ± 0.1
87.7 ± 2.0 2.4 ± 0.0
85.0 ± 0.3 NDt
84.4 ± 0.8 1.2 ± 0.0
* Defined as
Particle no. in presence of Con A
1-
_ Particle no. in presence of Con A 1
tMean ± standard deviation.
tNot determined.
Alteration of Cell Surface Properties by
Vesicle Adsorption
In the foregoing studies (Huang,
Ozato, and Pagano) on the mechanism
of lecithin vesicle uptake by lympho-
cytes, it was shown that DML or DPL,
but not EYL or DOL, vesicles become
associated with lymphocytes princi-
pally through a stable vesicle-to-cell
adsorption process. Thus, those lipid
vesicles which have the most signifi-
cant effect on the mitogenic response
and on lectin binding and agglutina-
tion are those which adsorb to the cell
surface. This correlation suggests that
the alteration of lymphocyte surface
properties may be a direct consequence
of vesicle-to-cell adsorption.
The stable adsorption of vesicles to
cells could lead to an enhanced binding
of mitogen to vesicle-treated cells, in
those cases in which mitogen binds di-
rectly to lipid. Thus, the enhanced Con
A binding observed in DML or DPL
pretreated cells may be a result of di-
rect interaction between Con A and ad-
sorbed vesicles. The finding that a sig-
nificant fraction of Con A binding is
noninhibitable by «-MM, particularly
in DPL treated cells, supports the idea
that some fraction of the binding might
represent a direct association of lipid
with Con A. In further support of this
notion is a recent study (van der Bosch
and McConnell, Proc. Nat. Acad. Sci.
U.S.A., 72, 4409-4413, 1975) demon-
strating a strong interaction of Con A
with DPL vesicles, which results in
vesicle- vesicle fusion.
The adsorption of vesicles to cells, as
well as the postulated vesicle-lectin in-
teractions, can also be used to explain
the differential effects of vesicle treat-
ments on lectin-induced agglutinabil-
ity. Thus, in the presence of Con A,
agglutination could be mediated both
by cell-bound Con A and Con A bound
to cell-associated vesicles. In this re-
gard it is interesting to note that the
agglutination of DPL or DML vesicle
pretreated thymocytes could not be
fully reversed by a-MM (Table 8). This
DEPARTMENT OF EMBRYOLOGY
61
may be due to the large residual Con A
binding found in cells treated with
these vesicles (Fig. 37).
The present findings demonstrate an
enhanced proliferation in stimulated T
cells and B cells following incubation
with various unilamellar lipid vesicles.
This observation is particularly signif-
icant, since T and B cells respond to
different specific mitogens and may
have different triggering mechanisms.
The stable adsorption of lipid vesicles
to lymphocytes may provide a common
element by which the enhancement in
these different cell types can take
place. Thus, the enhancement of cell
proliferation following vesicle treat-
ment may be due to the increased
mitogen binding resulting from
vesicle-to-cell adsorption and the sub-
sequent utilization of mitogen in
vesicle-treated cells.
In conclusion, our results suggest a
direct correlation between the surface
localization of exogenous lipids result-
ing from vesicle to cell adsorption and
their ability to modify lymphocyte cell
surface properties such as agglutina-
bility and lectin binding. It is further
suggested that vesicle-to-cell adsorp-
tion can have profound physiological
consequences in treated cells — namely,
enhancement of the proliferative activ-
ity of stimulated lymphocytes.
Studies of Membrane Adhesion
M. Takeichi and R. E. Pagano
with the assistance ofW. Duncan
The adhesive properties of cells are
among the more important factors con-
trolling cell behavior and mor-
phogenesis in multicellular organisms.
While many investigators have at-
tempted to elucidate the mechanisms
of cell adhesion, the molecular basis of
this phenomenon remains unclear. In
4-
2-
I I Total Binding
Binding Inhibitable by a _ MM
DML
DPL
EYDChol
I'l
EYL
DOL
Cont.
Fig. 37. Effect of different lipid vesicle pretreatments on the binding (1 hr at 2°C) of [ 125 l] Con A ( 1
fig/ml) to normal thymocytes.
62
CARNEGIE INSTITUTION
the present study of membrane-
membrane interactions, we describe
the adhesive properties of Chinese
hamster V79 fibroblasts to each other
as well as to artificial lipid vesicles,
and we present a preliminary charac-
terization of some of the cell surface
molecules that may play a role in the
adhesion process.
Vesicle to Cell Adhesion
Elsewhere in this report we have
documented the formation of a stable
adsorption between unilamellar phos-
phatidyl choline vesicles comprised of
dimyristoyl lecithin (DML) or dipal-
mitoyl lecithin (DPL) in vesicle-treated
lymphocytes. Such a stable vesicle-to-
cell adhesion appears to be a general
phenomenon and has also been found
to occur in the V79 fibroblast cell line.
The characteristics of this vesicle-to-
cell adhesion phenomenon are de-
scribed below.
Monolayer cultures of V79 cells were
dissociated into single cells with
chelating agents (EDTA; E cells) or a
combination of EDTA followed by tryp-
sin treatment (ET cells). The dispersed
cells were then incubated with [ 3 H]
DML vesicles for 1 hr at 2 °C, washed
and analyzed for the incorporation of
radioactive phospholipid. Figure 38
shows the dependence of DML uptake
on vesicle concentration. Between con-
centrations of 0.005 and 5.0 mg/ml of
applied DML vesicles, the DML uptake
increased monotonically from about 0.1
to 8.7 x 10 9 molec DML/cell. Further-
more, the technique used to dissociate
the cells from the culture dish prior to
vesicle-cell incubation had a marked
effect on the amount of DML which be-
came cell associated, with E cells tak-
ing up about an order of magnitude
9.0
<u 5.0-
Q 4.0
E 3.0
0.00
10.0
(ng. DML/cell) applied
Fig. 38. Concentration dependence of DML uptake by V79 cells in suspension. Cells were dis-
sociated with EDTA (o o) and subsequently treated with trypsin (x x) prior to incubation with
vesicles.
DEPARTMENT OF EMBRYOLOGY
63
more exogenous phospholipid than ET
cells. The presence of the divalent cat-
ions (Ca ++ or Mg ++ ) in the vesicle sus-
pension had no effect on the amount of
DML uptake in E or ET cells.
EM autoradiograms of E cells and
ET cells (Fig. 39) treated with [ 3 H]
DML vesicles (1 mg/ml) for 1 hr at 2 °C
clearly show that the radiolabeled
DML is confined primarily to the cell
surface, consistent with the stable
adhesion of this vesicle type of cell. Ad-
ditionally, it is seen that the amount of
radiolabel found associated with ET
cells is considerably less than for E
cells, in agreement with the data in
Fig. 38. In preliminary experiments, a
protein homogenate of E and ET cells
incubated with [ 3 H] DML vesicles was
subjected to SDS-polyacrylamide gel
electrophoresis. The distribution of
[ 3 H] DML in the gel suggests that a
few of the protein bands had a high af-
finity for the radiolabeled DML. Thus,
it is concluded that the adsorption of
intact DML vesicles may involve some
trypsin-sensitive material at the cell
surface. A further characterization of
this material is in progress.
Cell to Cell Adhesion
Monolayers of V79 cells were dis-
sociated into single cells with chelating
agents or proteases, and the ability of
these cells to reaggregate was deter-
mined. Cells dispersed with EDTA (E
cells) possess strong reaggregability in
the absence of divalent cations, which
was enhanced in the presence of CA 2+
(Fig. 40a). Cells dispersed with crystal-
line trypsin in Ca 2+ - and Mg 2+ -free
medium or with trypsin in the presence
of EDTA (TE cells) had no ability to
reaggregate in any medium tested
(Fig. 40c). Cells dispersed with crystal-
line trypsin in the presence of Ca 2+ (TC
cells), however, totally retained the
reaggregability in a Ca 2+ containing
medium but showed no indication
that there may be two pathways for
cell-to-cell aggregation, a Ca 2 + -
independent and a Ca 2+ -dependent one.
The function of Ca 2+ -dependent adhe-
sion sites seems to be protected from
trypsin digestion by the presence of
Ca 2+ (Fig. 40b vs. 40c). Mg* + at physio-
logical concentration could not substi-
¥
Fig. 39. EM autoradiographs of V79 cells incubated with [ :! H] DML vesicles (1 hr at 2°C). (a)
EDTA-dissociated cell, (b) EDTA-dissociated and trypsin-treated cell.
64
CARNEGIE INSTITUTION
Nt
No
1.0
— O O Q J
-Ca
5 9 e
-Ca or + Ca
— ©■ —
— ©
0.8
- \
Nq -Ca
0.6
N. °
0.4
\ + Ca
-
0.2
a
+ Ca ^^^
i i
b
i i i
» C
i
30 45
30 45
M I N
15 30
45
Fig. 40. Reaggregation of cells dissociated with 1 mM EDTA (a), 0.01% trypsin + 0.1 mM Ca 2+ (b),
and 0.017r trypsin + 1 mM EDTA (c) in the medium with or without 1 mM Ca 2 + . The extent of
aggregation was represented by an index Nt/No, where No is the initial particle number in cell
suspension and Nt is the total particle number at the time t,. of incubation.
tute for Ca 2+ in the phenomenon de-
scribed above.
Additional experiments were carried
out to compare the adhesive properties
of cells to a noncellular substrate with
the cell-to-cell adhesion described
above. Ca 2+ (but not Mg 24 " or Mn 2+ ) was
totally ineffective in promoting the
adhesion of cells to gelatin-coated plas-
tic substrate. Furthermore, reagents
that inhibit cell-to-substrate adhesion
(e.g., cytochalasin B, colcemid, and tet-
racaine) did not significantly affect the
Ca 2+ -dependent cell-to-cell adhesion.
TC and TE cells were not markedly dif-
ferent in their adhesiveness to noncell-
ular substrates. These observations
suggest that a Ca 2+ -dependent mecha-
nism observed for cell-to-cell adhesion
is absent in cell-to-substrate adhesion.
In order to explore the possibility
that changes in cell-to-cell adhesive-
ness correlate with modifications of cell
surface proteins, the surface proteins of
intact fibroblasts were labeled with 125 I
using lactoperoxidase, and the protein
species subsequently examined by
SDS-polyacrylamide gel electro-
phoresis. While a number of proteins
are labeled with 125 I in all three cell
types (Fig. 41a), several iodinated
bands were more intensely labeled in E
cells than the others, suggesting that
one of the cell surface proteins could be
related to the Ca 2+ -independent adhe-
siveness of cells. The difference in iodi-
nated bands between TC and TE cells,
however, was clear and simple. One
protein whose molecular weight (MW)
is in a range of 140K to 150K was pre-
dominantly found in TC cells. No other
qualitative difference in iodinated
bands was observed between these two
types of cells. The above 140-150 MW
band was also present in E cells.
The correlation of the presence of
140-150 MW protein with the ability
of cells to aggregate in Ca 2+ was also
found in another experiment. TE cells
did recover the Ca 2+ -dependent adhe-
siveness after several hours of incuba-
tion in normal cell culture medium.
This recovery could be completely in-
hibited by the addition of cy-
cloheximide to the medium. Interest-
ingly, the recovery of cell adhesiveness
DEPARTMENT OF EMBRYOLOGY
a b m
65
ft
1 2 3
Fig. 41. Autoradiograph of [ 125 I]-labled cell-
surface proteins electrophoresed in 7.57c SDS-
polyacrylamide gel. (a) E cells (1), TC cells (2),
and TE cells (3). (b) TE cells cultured for 6 hr in
the medium without (1) and with (2) 100 /xg/ml
cycloheximide. Iodination was carried out in
Ca 2+ -free medium at 5°C (a) and at 20°C (b). The
difference in the pattern of iodinated bands be-
tween (a) and (b) is due to the different tempera-
ture used for the iodination. Bars indicate the
position of 140-150K molecular weight region.
Top is cathode.
directly correlated with the appearance
of the 140-150 MW protein at the cell
surface of TE cells (Fig. 41b).
Evidence was also obtained suggest-
ing that the 140-150 MW cell-surface
protein can interact with Ca 2+ . This is
supported by the finding that the
radiolabeling of this protein was sensi-
tive to the presence of Ca in the iodina-
tion medium. Thus, the 140-150 MW
protein in TC cells could be iodinated
only in the absence of Ca 2+ (Fig. 42a vs.
42c), while the labeling of the other
proteins in TC cells and all the proteins
in TE cells were unaffected by the pre-
sence or absence of Ca 2+ (Fig. 42b).
This effect of Ca 2+ might be equivalent
to another effect of Ca 2+ to protect the
140-150 MW protein from trypsin di-
gestion. These phenomena suggest that
the 140-150 MW protein undergoes a
CM MIGRATED
Fig. 42. Densitometer traces of autoradio-
graphs of [ ,25 l] -labeled cell-surface proteins elec-
trophoresed in 7.5% SDS-polyacrylamide gel. (a)
TC cells iodinated in the absence of Ca 2 + ; (b) TE
cells iodinated in the absence or presence of Ca 2+ ;
(c) TC cells iodinated in the presence of Ca 2+ . Bar
indicates the position of 140-150K molecular
weight region in each figure.
conformational change or alters its
molecular state in the cell membrane
as a result of an interaction with Ca 2+ .
In summary, these observations
suggest a correlation of the appearance
of the 140-150 MW surface protein
with the Ca 2+ -dependent mechanism
for cell adhesion. The ability of this
protein to interact with Ca 2+ is consis-
tent with the possible participation of
this protein in a Ca 2+ -dependent cell
adhesion mechanism. Future experi-
66
CARNEGIE INSTITUTION
ments will focus on the isolation and
purification of this protein to deter-
mine its role in cellular adhesion pro-
cesses.
Distribution of Lipophilic
Fluorescent Probes in Intact
Cells
R. E. Pagano, K. Ozato, and J. M. Ruysschaert*
The use of small lipophilic fluores-
cent molecules for probing the physical
state of lipid bilayer model membranes
has been the subject of investigation by
"membranologists" for a number of
years. Such probes partition into the
hydrophobic phase of the lipid bilayer
and can be used to detect, by fluores-
cence changes, subtle changes in the
molecular packing of the lipid
molecules in the vicinity of the probe.
Recently, such probes have been used
to monitor the membranes of intact liv-
ing eukaryotic cells; however, no direct
examination of the location of these
molecules in whole cells has been
made. We report here on the in-
tracellular distribution of three such
fluorescent probes, l,6-phenyl-l,3,5-
hexatriene (DPH), perylene, and 2-p-
toluidinyl-6-naphthalene sulfonate
(TNS), in murine thymocytes, using
radiolabeled probes and the technique
of EM autoradiography.
DPH, perylene and TNS were
radiolabeled by tritium gas exposure
method of Wilzbach and subsequently
purified by crystallization from the ap-
propriate organic solvents. Labeling of
mouse thymocytes was performed with
a 2 x 10" 6 M solution of each of the
probe molecules dispersed in 0.15 M
KC1. One volume of cell suspension
(0.5-2 x 10 7 cells/ml) was mixed with
one volume of probe suspension and in-
cubated for 60 min at 25 °C. The
labeled cells were then washed twice,
fixed in glutaraldehyde, and processed
for electron microscope autoradiog-
raphy by standard methods.
*Free University of Brussels.
Micrographs of thymocytes treated
with [ 3 H] fluorescent probes were
taken at 14,000 x , and the outline of in-
dividual cells and location of grains
traced on paper. Each traced cell was
then divided into two compartments
and the number of silver grains in each
compartment determined. A cell-
surface compartment was defined by a
zone two half distances (HD) wide and
centered over the cell-surface mem-
brane contour. The remaining portion
of the cell defined the internal com-
partment. The HD determined by
Salpeter for the electron microscope
autoradiographic conditions used in
the present study ( :i H; 100-nm sections;
Ilford L4 purple interference color;
Microdol-X development) was 160 nm.
For each cell, the areas of these com-
partments were measured. Grain den-
sities were obtained by dividing the
number of grains in each compartment
by the area of that compartment. The
ratio of surface grain density to inte-
rior grain density and the average
±SEM were then calculated. The re-
sults of these measurements for each of
the probes tested are summarized in
Table 9. It is readily seen, both from
the percentage of silver grains found in
the surface compartment and from the
ratio of surface to internal compart-
ment grain densities, that in DPH and
perylene the probes are largely inter-
nalized in the treated cells while most
TNS is confined to the cell surface.
The cell population used for these
studies was contaminated with a small
fraction ( — 1%) of red blood cells and
thus permitted a simultaneous
analysis of the distribution of each of
the probes in this cell type. The data
obtained for [ :! H] DPH is given in Fig.
43 and shows a symmetrical distribu-
tion of the radioactivity about the cell
membrane, with approximately 80% of
the grains found within a distance of ±
0.3 microns of the membrane surface.
Similar data were obtained for [ 3 H]
TNS and [ 3 H] perylene.
These autoradiographic findings are
consistent with the known physical
DEPARTMENT OF EMBRYOLOGY
67
TABLE 9. Distribution of Autoradiographic Grains over Mouse Thymocytes
Treated with [ :i H] Fluorescent Probes
Probe
(Surface Grain
Percent Grains
Total Grains in Surface (Interior Grain
Counted Compartment Density)
± S.E.M.
! H TNS
! H DPH
! H Penylene
302
990
918
69
16
27
11.4 ± 1.3
2.0 ± 0.3
2.4 ± 0.1
o 10-
II
-0.7 -0.6 -0.5 -0.4 -0.3
-0.2 -0.1 +0.1 +0.2
Distance (microns)
1-0.3 +0.4 +0.5 +0.6 +0.7
Fig. 43. Distribution of autoradiographic silver grains in murine red blood cells, treated with [ 3 H]
DPH. Number of grains found is plotted vs. their distance in microns from the cell surface ( + , outside;
-, inside).
properties of the fluorescent probes
used, and suggest that except for TNS
or cells such as the erythrocyte, consid-
erable ambiguity may arise in inter-
preting fluorescence data obtained
with whole cells containing various in-
tracellular membranes into which
these probes can partition.
MEMBRANE TRANSPLANTATION BY FUSION
L. Huang and A. K. Ritchie
The goal of this work is to establish
methods by which one can transfer a
membranous protein molecule with a
well-defined function to the plasma
membrane of cells normally devoid of
such function ("membrane transplan-
tation") and to study the physiological
function of the protein in its altered
environment.
Since most hydrophobic proteins re-
quire a lipid environment for their
function, the transplantation process
can best be accomplished by using the
lipid membrane as a carrier for the pro-
68
CARNEGIE INSTITUTION
tein. Joining the carrier membrane
with the receiving membrane, a fusion
process, should then transfer the pro-
tein molecules to their target mem-
brane. One approach is to solubilize
and isolate protein molecules and to re-
constitute them into lipid vesicles. The
reconstituted vesicles can then be used
for fusion with the recipient cell. An
alternative method is to fuse intact
membrane vesicles, isolated from a
source very much enriched with a pro-
tein, directly with the plasma mem-
brane of the recipient cell. The
molecule used in this study is the
nicotinic acetylcholine receptor
(AChR), an integral membrane protein
responsible for the selective ion per-
meability found in the postsynaptic
membrane of some nerve and muscle
cells. We hope to introduce chemosen-
sitivity toward cholinergic agonists in
recipient cells. The ACh receptor is a
convenient protein, since in addition to
its well-defined functional activity it
can also be tagged in a highly specific
and relatively irreversible manner
with 125 I-a-bungarotoxin.
We have isolated AChR from the
electroplax of Torpedo californica. The
electroplax contains membrane that is
extremely rich in nicotinic acetyl-
choline receptors. Tissue from the elec-
troplax (stored at -70 °C) was
homogenized in a buffered isotonic salt
solution, centrifuged (100,00G£ x 30
min at 4°C), then successively washed
in a high and low salt concentration be-
fore solubilization by 2% triton X-100
(1 hr at room temperature). All solu-
tions contained 0.02% NaN :i and 5 mM
EDTA to minimize bacterial contami-
nation and protein degradation and
were kept at 4°C where possible. Par-
tial purification and detergent ex-
change were achieved by sedimenting
the AChR into a sucrose density gra-
dient containing 1% Na cholate. AChR
sedimented as a 10S peak, occasionally
with faster sedimenting shoulder. This
preparation had about 0.2 /umoles of
a-bungarotoxin binding sites/g protein
which is about 20-fold less than that
reported for Torpedo AChR purified by
affinity chromatography. The partially
purified AChR was then mixed with
dioleyl lecithin (DOL) which had been
predispersed in 2% Na cholate solu-
tion. The final concentration of lipid
was 10 mg lipid/ml in 1.5% cholate, 100
mM NaCl, 5 mM EDTA, 0.02% NaN 3
and 10 mM Tris, pH 7.4. The clear
mixture was then exhaustively
dialyzed against six changes of the buf-
fered salt solution at 4°C for 72 hours.
The solution became slightly turbid
after dialysis. Thin-section transmis-
sion electron microscopy revealed the
presence of very uniform unilamellar
vesicles with an average diameter of
500 A (Fig. 44a). If the DOL-cholate
dispersed solution was dialyzed with-
out protein, similar vesicles of smaller
size (average diameter 400 A) were
found (Fig. 44b). The homogeneity and
efficiency of vesicle formation was also
demonstrated by the presence of a
single peak on Sepharose 4B gel
chromatography which represented
99% of the radioactivity when using
tritium labeled DOL. The reconstitu-
tion of the AChR into vesicles was de-
termined by velocity sedimentation in
a detergent free sucrose density gra-
dient. AChR was always associated
with the 3 H-DOL vesicles which floated
at the top of the gradient (Fig. 45a). If
vesicles were first ruptured by 2% tri-
ton X-100, dissociation of AChR from
the lipids occurred (Fig. 45b) with the
lipids remaining at the top of the gra-
dient and receptor migrating to its
usual 10S position. The amount of re-
ceptor which could be incorporated into
the vesicle fraction increased linearly
then saturated as the protein/lipid
ratio reached 45 ng protein/mg lipid.
About two a-bungarotoxin binding
sites per vesicle were found under sat-
uration conditions. Such a low satura-
tion level of reconstitution may be due
to the differential affinity of AChR and
other contaminating proteins for the
hydrophobic sites in the lipid vesicles.
We expect much higher efficiency for
reconstitution when pure AChR is
used. The reconstituted vesicles were
also tested for 125 I-a-bungarotoxin ac-
DEPARTMENT OF EMBRYOLOGY
69
■r "*0,.
* it* >
.'S '<•.
b
Wr^ .
• >#
.¥.
m
3
m
,i s " pe
*£t ■
Fig. 44. Reconstituted vesicles prepared by dialysis method, (a) Vesicles containing AChR and
DOL. (b) Vesicles containing DOL alone. Bar is 2,000A.
16 20 4 8 12
Fraction number
Fig. 45. Velocity sedimentation profiles of '-^I-a-bungarotoxin labeled AChR and 'H-DOL in the
reconstituted vesicles in (a) detergent-free linear sucrose density gradients (5-20%- w/w), and (b)
similar gradient but containing 1% triton X-100. Fractions of 0.22 ml were collected from the bottom
of the gradient. These receptors were prelabeled by treatment of the crude triton extract with 0.1
/xg/ml '- s I-a-bungarotoxin prior to reconstitution. Free ,25 I-a-bungarotoxin was separated from toxin
receptor complexes on a BioGel P60 column.
70
CARNEGIE INSTITUTION
tivity in the presence or absence of tri-
ton X-100. About 80% of the total
AChR in the vesicle population was
available for binding when vesicles
were not ruptured by detergent, i.e.,
most of the reconstituted receptors had
their toxin-binding sites facing the ex-
terior of the vesicle.
Simple adsorption of the receptor to
the exterior of the lipid vesicle seems
unlikely, since co-migration of lipid
and receptor on a triton-free sucrose
gradient failed to occur when dialyzed
receptor was mixed with preformed
vesicles. We have also attempted to
distinguish between incorporated re-
ceptor and loosely adsorbed receptor by
testing for sensitivity to chymotrypsin
digestion. Chymotrypsin digestion (0.1
mg/ml) of receptor extracted with tri-
ton from Torpedo or from cultured
chick myotubes and of dialyzed Tor-
pedo receptor results in a shift of 10S
material to a 3S position on the sucrose
gradient. On the other hand, receptors
on the surface of cultured chick
skeletal myotubes or on reconstituted
vesicles remain as 10S material after
digestion by chymotrypsin. Thus, re-
ceptors associated with vesicles occupy
a protected position (i.e., failure to be
degraded to 3S material) similar to the
receptors in the membrane of intact
chick myotubes. The comparable ex-
periments on Torpedo membranes are
still in progress. These results encour-
age us to believe that the reconstituted
receptor is an integral part of the vesi-
cle membrane and not merely adsorbed
to its surface.
When the reconstituted AChR vesi-
cles were incubated with Chinese
hamster V79 cells for 45 min at 37 °C,
extensive uptake of AChR by the cells
was found with no evidence of
cytotoxicity. The intact cell-associated
AChR remained available for specific
125 I-a-bungarotoxin (the binding was
inhibited by 10~ 4 M curare and formed
a receptor toxin complex that migrated
as 10S on sucrose gradients). The
vesicle-treated cells acquired about 5x
10 4 a-bungarotoxin binding sites per
cell. An autoradiograph of a sparsely
plated culture dish of V79 cells that
had been treated with reconstituted
vesicles then labeled with 125 I-oi-
bungarotoxin revealed that most of the
grains were directly over cells. This is
in contrast to the random labeling of
the dish when cells were treated with
dialyzed receptors. An EM autoradio-
graph of vesicle-treated cells that had
been labeled with 125 I-a-bungarotoxin
further showed that the 125 I-label was
randomly distributed over the cell sur-
face. The persistence of the surface lo-
cation of AChR in these cells was also
examined. Vesicle-treated cells that
were incubated for various times in
complete growth medium at 37 °C still
retained over 70% of their original ac-
tivity when tested for toxin binding
after 10 hr of incubation. If vesicle-
treated cells were labeled with 125 I-a-
bungarotoxin and subsequently incu-
bated at 37 °C, those cells also retained
over 70% of their radioactivity after 10
hr (Fig. 46). About 43% of the radioac-
Fig. 46. The persistence of labeled receptor on
the surface of V79 cells containing transplanted
AChR. Monolayer cultures of V79 cells were in-
cubated for 45 min at 37°C with reconstituted
AChR. The cells were washed three times with
Gey's medium, then labeled with ,25 I-a-
bungarotoxin. The washed cultures were then
incubated at 37°C for the indicated amount of
time. The percentage of '-"'I-a-bungarotoxin re-
maining on the culture is plotted against the in-
cubation time.
DEPARTMENT OF EMBRYOLOGY
71
tivity that was lost from the cells into
the medium after 10 hr of incubation
was receptor-toxin complex, while 57%
was low molecular weight material (in-
cluded on a Bio-Gel P60 column) simi-
lar to the receptor- 125 I-a-bungarotoxin
degradation product that results from
the metabolism of the internalized
membrane of cultured chick myotubes.
Thus, the specific ct-bungarotoxin bind-
ing activity of AChR can be trans-
planted to the surface of V79 cells pre-
viously devoid of any toxin-binding
ability, and such activity remains on
the cells for relatively long periods of
time.
The uptake of receptor by V79 cells
treated with vesicles was time depen-
dent (Fig. 47a) but independent of the
temperature of incubation. This is in
marked contrast to the uptake of
labeled lipid by the same cultures. The
uptake of labeled lipid was dependent
on both the time and temperature of
incubation and consistent with the
kinetics observed using vesicles formed
by other methods. The amount and na-
ture of lipid uptake was the same
whether pure lipid vesicles or vesicles
containing receptor were used. Only
0.1% of the receptor was associated
with the cells after 45 min of incuba-
tion at 0°C, whereas 0.33% of the
lipids were associated with the cells
under the same conditions (Table 10).
When the study was repeated with ves-
icles containing prelabeled receptors,
then the percentage of incorporation
was higher, since this assay measures
internalized receptors as well as the
externally available receptors. In this
case the percentage Of incorporation of
the lipid (0.26%) and! that of the recep-
tor (0.2%) at 0°CJwere nearly the
same. The discrepancy between the
amount of lipid uptake and the recep-
tor uptake becomes very large at 37 °
due to the absence of any enhancement
of receptor uptake at the higher tem-
perature. The mechanism responsible
for the temperature-dependent portion
of lipid uptake (a process involving ves-
icle fusion) is thus different from the
process involving receptor uptake. It is
not clear to us at this time how the re-
ceptor is being transplanted.
Fig. 47. The uptake of AChR and :! H-DOL by V79 cells. A confluent 35-mm dish of V79 cells was
incubated with reconstituted vesicles at 0° (• • ) and 37°C (O O) for the indicated times. Each
dish was treated with vesicles containing 8 mg/ml of DOL and 0.36 fig/ ml of protein, then washed
three times with Gey's medium. The plates were then incubated with 0.1 fig/ml of 125 I-a-bungarotoxin
for 30 min at room temperature. The cells were washed six times with Gey's medium, then extracted
with 1% triton and counted by liquid scintillation (for :i H; Fig. 47B) and/or a Gamma well counter (for
,25 I; Fig. 47A).
72
CARNEGIE INSTITUTION
TABLE 10. Uptake of Lipid and ACh Receptor by V79 Cells Treated with Vesicles
Containing the Reconstituted Receptor
Temp. (°C)
'H-DOL incorporation
into cells as percent
of the total :i H-DOL
applied
Receptor incorporation
into cells as percent
of the total applied
* Receptor and 3 H-DOL
vesicles
t 125 I-a-bungarotoxin
receptor and :! H-DOL
vesicles
37
37
0.33
1.1
0.26
0.66
0.1
0.1
0.2
0.2
* Cells were treated for 45 min at 37 °C with vesicles containing unlabeled receptors. The cultures
were then washed three times with Gey's medium and incubated for 30 min at room temperature with
0.1 /ti-g/ml 125 I-a-bungarotoxin. Cultures were washed 6x5 min, then extracted with 19c triton X-100.
tCells were treated in the same manner as above except that the vesicles contained receptors that
had been prelabeled with 125 I-a-bungarotoxin prior to reconstitution.
We have also attempted to introduce
the AChR into the plasma membrane
of V79 cells by treating these cells with
membrane fragments isolated from the
Torpedo electroplax. A membrane frac-
tion containing 10 2 moles of
a-bungarotoxin binding sites/g protein
was obtained by sedimenting a crude
membrane preparation on a discon-
tinuous sucrose density gradient.
Thin-section electron microscopy of
this fraction showed closed membrane
vesicles of very heterogeneous size dis-
tribution (0.1-2 /u. in diameter). A sus-
pension of this material was incubated
with a monolayer of cultured V79 cells
in the presence of UV-inactivated Sen-
dai virus or lipid vesicles comprised of
90% phosphatidyl choline and 10%
phosphatidyl glycerol or phosphatidyl
serine. Some toxin binding activity was
observed, but this method did not ap-
pear to be as promising as the use of
the reconstituted vesicles to transplant
membrane proteins.
Finally, we have tried to test our re-
constituted vesicles and V79 cells
treated with the reconstituted vesicles
for some indication of functional activ-
ity by the receptor (i.e., ion permeation
specifically induced by a cholinergic
agonist). The most satisfactory method
of testing the vesicles for 22 Na ion flux
was a 1-ml P60 Bio-Gel column in
which vesicle-associated 22 Na could be
completely separated from free 22 Na in
1 min. When reconstituted receptor
vesicles were tested, they demon-
strated an uptake of 22 Na which was
saturated within 10 min. This influx,
however, was not enhanced by the pre-
sence of carbachol, a cholinergic
agonist. Attempts to test the V79 cells
that had been treated with vesicles for
AChR activity by electrophysiological
techniques were also negative. These
cells had resting potentials of about
-25 mV but did not respond to bath or
iontophoretic application of ACh.
There was no change in the ACh sen-
sitivity of cultured chick myotubes
similarly treated with reconstituted
receptor after blockade of endogenous
receptors with a-bungarotoxin. There
are a number of possibilities for our
failure to detect receptor functional ac-
tivity: (1) The receptor, although re-
taining its oi-bungarotoxin binding ca-
pacity, may have lost its ability to
function as an ionophore due to some
denaturation or degradation of the pro-
tein. (2) The isolated receptor may be
lacking some other necessary compo-
nent that was discarded or inactivated
in the isolation process. (3) The recep-
tor may have been transferred but was
not situated in the membrane in an in-
tegral manner that allowed it to func-
tion as an ionophore. Limited success
by other investigators in restoring both
crude and purified AChR into artificial
DEPARTMENT OF EMBRYOLOGY
73
membranes and vesicles, although
with only intermittent success, leads
us to believe that the first possibility is
the most likely reason for our failure.
Experiments are in progress to de-
termine the conditions necessary for
restoring functional activity to the re-
ceptor. The mechanism by which
transplantation occurs will also be of
primary importance, using either the
ACh receptor as a marker — -provided
that functional activity can be
restored — or some other membrane
protein. Our preliminary data indicate
that this is a promising method of
membrane transplantation.
MECHANISM OF SYNGENEIC RESPONSE INDUCED BY LECTINS
K. Ozato and J. D. Ebert
In Year Book 74, pp. 90-92, we dem-
onstrated that the modification of the
lymphocyte surface by lectins renders
the cells responsive to syngeneic lym-
phocytes, as revealed by the prolifera-
tion of T cells stimulated by spleen
cells bearing the same histocompatibil-
ity antigens. A syngeneic response can
be obtained when either responding
cells or stimulating cells are pretreated
with native or succinyl-Con A (N-Con
A or S-Con A).
During the past year we attempted a
more analytical study of this syngeneic
response. Two directions of research
were undertaken. We first undertook
to determine the characteristics of the
cellular stimulus produced by
syngeneic spleen cells. We have ob-
tained data that suggest that the re-
sponse is characteristic of T cells. Con-
genital athymic mice (nu/nu) provide a
lymphocyte population deficient in T
cells. The spleen cells from nu/nu mice
did not respond to syngeneic cells.
Further, in an allogeneic situation,
they did not manifest the Con A pre-
treatment effect that usually augments
the mixed lymphocyte reaction. Con A
coating of either responding cells or
mitomycin-C treated spleen cells failed
to respond in nu/nu mice. Since we first
observed the syngeneic response in a
pure T-cell population, we now believe
that the response is a pure T-cell spe-
cific phenomenon.
The analysis of the stimulating
population has been carried out by
treating the cells with antibodies
against specific surface components of
T cells or B cells. Goat antimouse Ig or
AKR anti CBA# were used for elimina-
tion of B cells and T cells, respectively.
Antibody treatment was carried out be-
fore the cells were coated with Con A.
The effect of antibodies was studied
when both responding cells and
stimulating cells were coated with Con
A. As shown in Table 11 (a and b) when
stimulating spleen cells were treated
with anti-Ig, the response was effec-
tively suppressed, whereas anti-0 anti-
body treatment did not affect the re-
sponse. This finding indicates that
stimulatory function depends on the
presence of B cells. We have tested the
possible participation of adherent cells
(mostly macrophages). Spleen cell
populations that are deleted of adhe-
rent cells by passing the cells through
plastic petri plates still manifest their
stimulatory function.
Then how do responding T cells
interact with stimulating cells (proba-
bly B cells)? Do B cells release a factor
into the medium? Do stimulating cells
somehow enable Con A molecules to
associate with T cells more effectively?
Stimulating and responding cells must
either be in contact or be "very close
neighbors," for the culture fluid in
which stimulating cells have been
grown does not stimulate. Moreover,
when stimulating cells and responding
cells are cultured separately, using
nucleopore filter membranes, the
syngeneic response is not observed.
The role of stimulating cells does not
appear to be to provide physical sup-
port of Con A molecules on the surface,
74
CARNEGIE INSTITUTION
TABLE 11(a). Effect of Anti-0 or Anti-Ig on Con A-induced Syngeneic Response
CRT vs. [N-Con A]-Syngeneic Spleen
3 H-TdR Uptake on Day
3 (cpm x 10 :i /culture)
Treatment
B10.D2
CBA
Medium RPMI
27.2 ± 3.0
18.3 ± 1.4
Normal Serum AKR (1/20)
30.3 ± 2.4
16.7. ± 1.0
Normal Serum Goat (1/20)
32.4 ± 2.6
16.2 ± 1.3
Rabbit C (1/15)
15.6 ± 1.6
17.6 ± 1.5
AKR Anti- 0(1/20)
25.3 ± 3.0
18.8 ± 1.0
AKR Anti- 0(1/40)
38.0 ± 2.5
16.0 ± 0.7
AKR Anti- 0(1/40) + C (1/15)
29.3 ± 3.0
17.2 ± 2.0
Goat Anti-Ig (1/20)
0.8 ± 0.1
0.8 ± 0.1
Goat Anti-Ig (1/40)
0.2 ± 0.0
1.0 ± 0.0
Goat Anti-Ig (1/40) + C (1/15)
0.4 ± 0.0
0.3 ± 0.0
Spleen cells pretreated with mitomycin C were then treated with antibodies for 45 min at 37°C in
the presence or absence of complement (C). The parentheses indicate the dilution of the antiserum.
After washing the cells to remove killed cells as well as unbound antibodies, cells were further treated
with N-Con A at 50 /u-g/ml at 37°C for 30 min. They were then mixed with responding cells
(cortisone-resistant thymocytes) in a 1:1 ratio (2x 10 6 cells) in 0.5 ml. The response was measured by
:l H-TdR uptake on day 3.
TABLE 1Kb). [N-Con A-CRT] vs. Syngeneic Spleen
! H-TdR Uptake on Day 3
(cpm x 10 3 /culture)
Treatment
C57BL/10
CBA
Medium RPMI
Normal Serum AKR (1/20)
Normal Serum Goat (1/15)
Rabbit C (1/15)
12.7 ± 0.5
10.3 ± 0.8
14.4 ± 1.1
13.0 ± 0.6
12.3 ± 0.7
11.5 ± 0.8
10.0 ± 0.8
10.0 ± 0.8
AKRAnti-0(l/2O)
AKR Anti- 0(1/40)
AKR Anti- 0(1/40) + C (1/15)
11.4 ± 0.7
10.9 ± 0.3
12.0 ± 1.0
9.0 ± 0.1
13.5 ± 1.3
8.9 ± 0.9
Goat Anti-Ig (1/20)
Goat Anti-Ig (1/40)
Goat Anti-Ig ( 1/40) + C ( 1/15)
0.3 ± 0.0
0.9 ± 0.1
0.2 ± 0.0
4.3 ± 0.7
4.7 ± 0.3
3.0 ± 0.1
Stimulating spleen cells were treated with antibodies in the same manner as (a), after which they
were mixed with responding cells that had been treated with N-Con A (50 /u.g/ml).
for when stimulating cells are killed by-
fixation or by freezing and thawing
after they were coated with Con A, the
response is completely abrogated (Ta-
ble 12). However, treatment with
metabolic inhibitors or the microfila-
ment disrupting agent, cytocalasin B,
did not affect the response. Thus we
may suggest that the intact cell mem-
brane of stimulating cells is required.
It is an interesting possibility that
membrane components may be ex-
changed between responding cells and
stimulating cells and that Con A
molecules are transferred from
stimulating cells to responding cells or
vice versa. We are now studying the
fate of Con A molecules, using 125 I-Con
A bound to either stimulating cells or
responding cells to test this possibility.
Another goal of our research has
been to delineate the role of lectins. If
the response is entirely dependent on
stimulating cells, the role of Con A or
other lectins must be to enhance the
"physical contact" of responding and
DEPARTMENT OF EMBRYOLOGY
75
TABLE 12. No Syngeneic Response from Fixed "Stimulating" Cells
! H-TdR Uptake on Day 3 (cpm xlO" 3 ")
Treatment of Stimulating Cells
I. Responding Cells:
Coated with N-Con A
II. Stimulating Cells:
Coated with N-Con A
1. No treatment
2. Glutaraldehyde(2.5%)
3. Ethanol
4. Acetone
5. Freezing and thawing
6. Cytochalasin B (40 fxg/ml)
7. Dinitrophenol (2 mM)
8. Sodium azide (50 mM)
18.5
0.2
0.3
0.1
0.2
23.7
22.4
23.1
15.6
0.0
0.3
0.1
0.1
18.4
19.2
17.8
Stimulating cells (mitomycin-C treated) were treated with various reagents for 15 min at room
temperature (2 to 4) or treated for 45 min at 37°C (5 to 8). Freezing and thawing were repeated twice.
In column II, Con-A pretreated cells were used. The response of cortisone-resistant thymocytes was
measured on day 3.
stimulating cells. If so, any lectin that
can agglutinate lymphocytes should
give the same effect. Among lectins we
have tested, similar syngeneic re-
sponse was observed with two other
lectins, namely, leucoagglutinin
(purified form of phytohemagglutinin)
(Fig. 48), and lentil lectin (a gift from
John Cebra at Johns Hopkins Univer-
sity). Although wheat germ aggluti-
nins agglutinate lymphocytes, they do
not induce the syngeneic response.
Binding to a sugar does not seem to be
a specific requirement. The ability to
15
1
1
i
h-
i
1 " " 1
10
1
1
i
I j
1 f
I |1
■1 f
1
~>
s
s
s
\
\
\
(
5
I 1
I j
i J
\
<
> I
~ *
*
* *
O 20
IOO 200 300 500
Leucoagglutinin Concentration (fiq/m\)
Fig. 48. CBA cortisone-resistant thymocytes were cultivated with syngeneic (CBA) (• •) or
allogeneic (B10.D2) (o o) spleen cells pretreated with leucoagglutinin at the concentrations indi-
cated at 37°C for 30 min. As controls, spleen cells (A A) were cultured alone.
76
produce a syngeneic response appears
to be limited to those lectins that have
mitogenic properties in T cells. It is
therefore conceivable that such lectins
CARNEGIE INSTITUTION
not only effect a physical association
between cells but also provide a
stimulus to the responding cells.
STUDIES ON SKELETAL MUSCLE AND NEURONAL PLASMA
MEMBRANES
D. M. Fambrough, R. E. Pagano, P. N. Devreotes, K. Tepperman, D. J. Card, and S. Carbonetto
with the technical assistance of A. Mabin, W. Duncan, andH. W. Chen
The acetylcholine (ACh) receptor is a
major glycoprotein in the muscle
plasma membrane of the neuromuscu-
lar junction and a minor component of
extrajunctional plasma membrane.
The ACh receptor has been our point of
focus in studies on the differentiation
and organization of skeletal muscle
plasma membrane, and it has become
clear that the number and distribution
of ACh receptors in skeletal muscle
(which vary with the developmental
and physiological state of the muscle)
are regulated largely by regulation of
ACh receptor metabolism. Our studies
of the metabolism of extrajunctional
ACh receptors have led us to propose a
model for the major events in receptor
metabolism, illustrated in Fig. 49.
The main features of the model are
as follows. The protein subunits of
which the ACh receptor is composed
are synthesized by the conventional
protein synthesizing mechanisms of
cells and are rapidly assembled into re-
ceptor units resembling those in the
plasma membrane. These newly syn-
thesized receptors occur upon some
membrane system(s) in the cytoplasm
of the muscle cells and are collectively
referred to as the "precursor pool." The
average residence time of a new recep-
tor in the precursor pool is 2-3 hr. Re-
ceptors are transferred from this pool
to the surface by an energy-requiring
process and are then able to function.
The average half-life of an ACh recep-
tor in the plasma membrane is 22 hr.
HYPOTHETICAL "LIFE-CYCLE" of ACETYLCHOLINE
RECEPTORS
mRNA(s)
V = PRECURSOR
V = SURFACE
M-
I-TYROSINE
Fig. 49. Hypothetical "life cycle" of acetylcholine receptors in cultured chick skeletal muscle. The
figure depicts a cross section through a myotube with receptors symbolized as triangles in membrane
profiles. Arrows indicate processes consistent with data on receptor metabolism. Symbols for precur-
sor and surface receptors and for 125 I-a-bungarotoxin and ,25 I-tyrosine are indicated on figure.
DEPARTMENT OF EMBRYOLOGY
77
Receptors are degraded by an energy-
requiring process that involves the in-
ternalization of the receptor and trans-
port to secondary lysosomes and then
proteolytic destruction.
This model for ACh receptor biosyn-
thesis and turnover was developed
mostly on the basis of data from exper-
iments involving the use of
a-bungarotoxin and [ 125 I] labeled
a-bungarotoxin as probes for ACh re-
ceptors. The strategies for using these
probes to measure the accumulation of
receptors in plasma membranes, the
incorporation of newly synthesized re-
ceptors into plasma membranes, and
the degradation of receptors have been
described in previous Year Books.
Some of the details of the models were
based upon data from studies involving
inhibitors of various cellular processes.
In Year Book 74 (pp. 68-73) we de-
scribed the application of two new
techniques to the study of ACh receptor
metabolism: electron microscope au-
toradiography and the "density shift"
technique. EM autoradiographic
studies during the past year have pro-
vided quantitative supporting data for
the degradative aspects of the model of
receptor metabolism. The density shift
technique has made it possible to label
ACh receptors directly and thus to ver-
ify the kinetics of receptor biosynthesis
and incorporation into plasma mem-
branes.
Now that we can confidently meas-
ure many steps in the metabolism of
extrajunctional ACh receptors, we
have begun to investigate the manner
in which external stimuli influence the
number and distribution of ACh recep-
tors in skeletal muscle fibers via their
influence upon receptor metabolism.
The two classes of stimuli postulated to
affect receptor metabolism are (1)
hypothetical "trophic" factors released
by motoneurons onto muscle fibers and
(2) muscle activity (which is normally
driven by the activity of motoneurons
but can be artificially driven by electri-
cal stimulating). So far we have only
investigated the effect of artificially
driven or suppressed muscle activity on
the metabolism of extrajunctional re-
ceptors. These studies suggest that
muscle activity leads to a reduction in
the number of receptors because of a
decrease in the rate of receptor biosyn-
thesis or incorporation into plasma
membranes.
The organization of ACh receptors
into dense clusters at postsynaptic sites
may involve active "gathering" of ex-
trajunctional receptors or merely
stabilization of receptors implanted in
the plasma membrane at the site of
nerve-muscle interactions. From an
experimental point of view, one can
pose the question: Can preexisting ex- '
trajunctional receptors be accumulated
into postsynaptic receptor arrays? To
investigate this question and also to
gain some insight into the number and
distribution of receptors on embryonic
skeletal muscle during differentiation
and synapse formation, we have begun
a study of the chick anterior latissimus
dorsi muscles. We are also developing
techniques to determine whether or not
ACh receptors on muscle are free to
move in the plasma membrane.
We consider the ACh receptor both a
functional entity of great interest in its
own right and also an example of a
membrane glycoprotein. In this latter
frame of reference, we are interested to
know whether the kinetics of turnover
of ACh receptors are typical of or iden-
tical to the kinetics of turnover of other
membrane proteins. To investigate this
matter we have been attempting to
label a set of membrane proteins by the
lactoperoxidase iodination technique
and then to study the turnover of this
class of proteins. Some progress and
the definition of some difficulties are
summarized in the report.
Our interest in excitable cells has ex-
tended to neurons, and as a first ap-
proach to studies of membrane
metabolism and control of membrane
properties in neurons we have selected
chick sympathetic neurons. This choice
was based largely upon the work of
Greene and collaborators, who have
demonstrated the specific binding of
a-bungarotoxin to such cells. We are
78
CARNEGIE INSTITUTION
currently testing whether or not
tv-bungarotoxin binds to ACh receptors
in cultures of such neurons and
whether the binding is strong enough
to permit us to develop strategies simi-
lar to those used to study the
metabolism of skeletal muscle ACh re-
ceptors.
ACh Receptor Biosynthesis
[ 125 I] mono-iodo-a-bungarotoxin. In
addition, a-bungarotoxin receptor
complexes can be sedimented through
dense sucrose-deuterium oxide gra-
dients. As will be demonstrated below,
receptors synthesized from [ 2 H, 13 C,
15 N] amino acids and complexed with
ct-bungarotoxin can be separated from
normal density cc-bungarotoxin-
receptor complexes on such gradients.
P. N. Devreotes
Purification of the plasma mem-
brane and its proteins poses a major
technical difficulty to membrane
biologists. The total plasma membrane
protein usually represents only about
1% of the cell protein, and specific
membrane components may represent
only a small fraction of this total. In
their investigations of membrane pro-
teins, therefore, membrane biologists
have turned to specific markers or spe-
cific chemical modification techniques
to tag these proteins. Marking methods
generally label membrane components
exposed on the outer surface of the
membrane after synthesis and after in-
sertion into the plasma membrane. Ac-
cordingly, attempts to study biosyn-
thesis and turnover of membrane com-
ponents have been indirect, including
our studies of ACh receptor metab-
olism with the use of ct-bungarotoxin.
Criticism is usually raised that the de-
gradation or turnover of the bound
marker rather than the native protein
itself is being studied. Attempts have
been made to study biosynthesis and
turnover directly by labeling the mem-
brane components with radioactive
amino acids. With this approach, how-
ever, the advantages gained through
the specificity of the marker must be
sacrificed. Marker specificity can be re-
tained in direct labeling experiments if
receptors are directly labeled not with
radioactive amino acids but with
amino acids that change a physical
property of the receptors — their densi-
ty.
A specific and nearly irreversible
marker for acetylcholine receptors is
['h, I2 c, ,4 n]
RECEPTORS
30
FRACTION
Fig. 50. Density-shifted receptors from dif-
ferentiating cells. Twenty large-muscle cultures
were grown for 24 hr in normal medium. At that
time, there were very few myotubes in culture,
and the myoblasts were fed with medium con-
taining [ 2 H, 13 C, 15 N] -amino acids. After the cells
had fused in the density-labeled medium, the
surface receptors were complexed with [ l25 l]
mono-iodo-a-bungarotoxin. The complexes were
then extracted into 259c cholate, 10-mM Tris and
mixed with a marker of [ i: "l] mono-iodo-a-
bungarotoxin-receptor complexes prepared from
myotubes grown in normal medium. A 0.4-ml
aliquot of the mixed extract was carefully
layered over a 25-40% sucrose-deuterium oxide
gradient containing Y7c triton, ImM EDTA-
PMSF, 10 mM Tris,pH 7.8. Centrifugation was
for 2 days in a Beckman SW41 rotor at 36K at
4°C. Gradients were pumped into scintillation
vials and the activity of each isotope determined.
DEPARTMENT OF EMBRYOLOGY
79
So far, the most reliable method of
preparing a sample of pure density
labeled [ 125 I] mono-iodo-a-bungaro-
toxin-receptor complexes is to allow
cells to differentiate in medium con-
taining [ 2 H, 13 C, 15 N] amino acids,
with [ 125 I] mono-iodo-a-bungarotoxin
and extracted in nonionic detergents.
Nearly complete separation of these re-
ceptors from [ X H, 12 C, 14 N]-receptors
has been obtained, as shown in Fig. 50.
Over 90% of the acetylcholine recep-
tors extracted from cells that have dif-
ferentiated in pH, 13 C, 15 N] amino acid
medium are of altered buoyant density.
An estimate of the apparent density
shift between PH, 12 C, 14 N]-receptors
and [ 2 H, 13 C, 15 N]-receptors can be ob-
tained by measuring sedimentation
rates of the respective complexes with
a-bungarotoxin in shallow gradients of
different average densities. The aver-
age density of the gradients was varied
by substitution of D 2 for H 2 0. It can
be shown that for shallow gradients the
logarithm of the distance traveled by a
given protein versus time is a straight
line. For a given protein, the slope of
this line is a function of the average
density of the gradient medium. In fact
if X is the slope, then
A = k (l-p lp)
where k is a constant dependent on
molecular weight, rotor speed, and
temperature; p is the density of the
protein; and p is the average density
of the medium. Plots of A. versus p ,
therefore, intersect the p axis at p.
This kind of analysis has been carried
out for PH, 12 C, 14 N]- and [ 2 H, 13 C,
15 N]-receptors. As shown in Fig. 51, the
apparent density of PH, 12 C, 14 N]-
receptors is 1.26 and that of [ 2 H, 13 C,
15 N]-receptors is 1.33. When correc-
tions are made for detergent binding to
the a-bungarotoxin receptor complexes
(detergent binding causes, in part, the
4 6 8 10
TIME (HOURS)
1.2 1.3
p (g/CM 3 )
Fig. 51. Estimate of density shift. Identical aliquots of the same mixed extract described in the
legend to Fig. 49 were layered onto identical gradients and centrifuged for different times. Gradients
for each series were 5-20% sucrose in H 2 0, 50% D 2 or D 2 containing 1% triton, 10 mM Tris.
Gradients were collected into scintillation vials, and the position of the two bands determined at each
time for the three different series. X„ and X refer to the distance from the center of the rotor to the
meniscus and the position of the band, respectively. Lambda is the slope of the straight lines shown on
the left. Experiment was repeated twice.
80
CARNEGIE INSTITUTION
anomalous low density of the receptor),
a density shift of 0.08 g/cm 3 is calcu-
lated. This corresponds to a shift of ap-
proximately 6%. The maximum at-
tainable shift (calculated from the
amino acid composition of purified re-
ceptors from Electrophorus electricus)
is about 8.5%. Since the calculated
shift is a minimum estimate, a sub-
stantial fraction of the amino acids of
the receptor must be substituted by
PH, 13 C, 15 N]-amino acids.
As has been described in the previ-
ous section, the model for acetylcholine
receptor biosynthesis suggests that re-
ceptors undergo a 3-hr transit time be-
fore being inserted into the plasma
membrane. This feature of the model
has been derived from observations
using puromycin and cycloheximide to
block protein synthesis. When protein
synthesis is inhibited, new receptors
are incorporated into the surface for
about 3 hr and then incorporation
stops, suggesting that the pool of pre-
synthesized receptor has been depleted.
Direct evidence of a 3-hr transit time
for receptors can be obtained by the
density shift technique. Density-
shifted [ 2 H, 13 C, 15 N]-receptors should
begin to appear in the cell surface
about 3 hr after their synthesis.
The experiment was carried out as
follows. The existing surface receptors
were first blocked by an overnight ex-
posure to unlabeled a-bungarotoxin.
After rinsing away the unbound
a-bungarotoxin, myotubes were incu-
bated in medium containing [ 2 H, 13 C,
15 N]-amino acids. New receptors then
appear on the surface of the myotubes
as a linear function of time (in these
experiments, the heavy medium may
have had a slightly deleterious effect
on the cultures at late times) and can
be labeled at subsequent times with
[ 125 I] mono-iodo-a-bungarotoxin. After
extraction into nonionic detergents,
these [ 125 I] mono-iodo-a-bungaro-
toxin-receptor complexes were mixed
with a marker of [ ,31 I] mono-iodo-a-
bungarotoxin-receptor complexes
extracted from myotubes grown in
normal medium. The mixed extract
was then centrifuged through
sucrose-deuterium oxide gradients to
determine the fraction of [ 2 H, 13 C,
15 N]-receptors. These density-shifted
receptors represent the fraction of
newly appearing receptors that were
synthesized after addition of dense
amino acids to the medium. As shown
in Fig. 52 new receptors begin to ap-
pear on the surface without a lag; how-
ever, [ 2 H, 13 C, 15 N]-labeled receptors
appear at the maximum rate only after
a lag of about 3.5 hr. Small amounts of
[ 2 H, 13 C, 15 N]-receptors appear at 2 and
3 hr; after 3.5 hr most of the newly ap-
pearing receptors are density shifted.
Since most receptors only appear on
the surface at the maximum rate 3.5 hr
after their biosynthesis, it is suggested
that newly synthesized receptors
undergo a 3-hr transit time within the
cell. Since small amounts of receptors
appear at 2 and 3 hr, however, it is
suggested that the intracellular trans-
port of receptors is not a strictly linear,
assembly-line process. Newly synthe-
sized receptors and receptors already in
transport can become, to a certain ex-
tent, intermixed and inserted at ran-
dom into the plasma membrane. Alter-
natively, the small amounts of receptor
appearing at 2 and 3 hr after synthesis
might be in a subpopulation of
myotubes in which the transit time is
shorter than average.
It has also been learned from these
experiments that acetylcholine recep-
tors are not "unmasked" during the de-
velopment of the cultures. Most of the
newly appearing receptors are synthe-
sized a few hours before incorporation
into plasma membranes. Furthermore,
as evident from the experiment de-
scribed in the legend to Fig. 50, the
acetylcholine receptors expressed dur-
ing differentiation of myoblasts into
multinucleated myotubes are newly
synthesized. It is unlikely, therefore,
that presynthesized receptors are
stored within myoblasts and expressed
during differentiation.
The density shift technique for
acetylcholine receptors has enabled us
to confirm several aspects of the model
DEPARTMENT OF EMBRYOLOGY
81
for acetylcholine receptor biosynthesis
without purification of the receptor or
the plasma membrane. We are using
density labeling technique to measure
the kinetics of incorporation of isotop-
ically substituted amino acids into the
receptors undergoing intracellular
transport and to measure directly the
half-life of receptors once they become
members of the surface population.
4 6
TIME (HOURS)
Fig. 52. Appearance of density-labeled receptors on surface membrane. Three-, four-, or five-day-old
cultures of myotubes were saturated with unlabeled a-bungarotoxin (0.2 /u,g/ml for about 15 hr). After
rinsing away unbound a-bungarotoxin, the medium was replaced with medium containing [ 2 H, I:i C,
15 N] amino acids. After incubation for the indicated times at 37°C, sets of cultures were cooled at 4°C
and saturated with [ 125 l] mono-iodo-a-bungarotoxin. The [ 125 l] mono-iodo-a-bungarotoxin-receptor
complexes were then extracted in 2.5% cholate, 10 mM Tris, mixed with a marker of 131 I mono-iodo-
a-bungarotoxin-receptor complexes prepared from myotubes grown in normal medium. Samples were
layered in thin bands on 10-20% sucrose-deuterium oxide gradients containing 1% triton, lOmM Tris.
After centrifugation for 1 day in a SW41 rotor at 36K at 4°C, gradients were collected into scintilla-
tion vials. Total receptors (o). Amount of material running to the position of [ 2 H, 13 C, 15 N]-
receptors (• ••) was determined by curve fitting. Different ratios of "heavy" and "light" peaks were
added together by a computer program which used, for the shape of each peak, the shape of the
marker on each gradient. These curves were matched to the data and could be fit within 5% . Similar
results could be obtained by counting the area outside the marker peak and expressing it as a fraction
of the total area.
Autoradiographic Analysis of
Receptor Degradation
P. N. Devreotes and D. M. Fambrough
The rate of appearance of radioactiv-
ity in the medium after [ 125 I]-mono-
iodo-a-bungarotoxin is bound to ACh
receptors is a measure of receptor de-
gradation. Our data on the mechanism
of degradation of receptors are consis-
tent with a model (Fig. 49) in which
receptors are internalized and then de-
graded. For example, homogenates of
myotubes labeled with [ 125 I] mono-
iodo-a-bungarotoxin have little or no
capacity for degradation; and deg-
radation is inhibited by 2,4-
dinitrophenol and is very temperature
sensitive, having a Q 1() of 8. After a
brief "pulse" incubation of myotubes
with [ 125 I] mono-iodo-a-bungarotoxin,
the rate of release of [ 125 I] mono-iodo-
tyrosine into the medium (due to deg-
radation of the [ 125 I] labeled
a-bungarotoxin initially bound to sur-
face ACh receptors) requires about 90
82
CARNEGIE INSTITUTION
min to reach a maximum (Fig. 53).
During this time some of the labeled
toxin-receptor complexes must move
into a small compartment of receptors
earmarked for degradation. Since the
lag time is equivalent to about one
hour of degradation and since receptors
are normally degraded at about 3% per
hour, the compartment should contain
about 3% as many receptors as there
are surface receptors. We have looked
for the subcellular location of this
hypothetical compartment, using elec-
tron microscopy autoradiography. In
these studies the location of radioactiv-
ity on and within cultured muscle cells
was determined as a function of time
after brief exposure of cells to [ 125 I]
a-bungarotoxin. The large secondary
lysosomes of the muscle cells were
found to accumulate radioactivity up to
a plateau level of slightly over 2% of
total cellular radioactivity (Fig. 53).
This accumulation occurred with kine-
tics consistent with a model of degrada-
tion in which the transport of [ 125 I]-a-
bungarotoxin-receptor complexes to
lysosomes and their hydrolysis in lyso-
somes results in the production of
[ 125 I]-tyrosine.
.08
.06
.04
.02
EARLY KINETICS OF
RECEPTOR DEGRADATION
[ l25 l] TYROSINE
IN MEDIUM
GRAINS IN
LYSOSOMES
<
a
Id
<
UJ
_l
UJ
K
UJ
> z
60 120
MINUTES AFTER CONTACT
0F[ l25 l] a-BUNGAROTOXIN
> C
O >
< h-
Q i
a ,
2 -
in
u. w
0<—
_ o
I-
o
<
Fig. 53. Kinetics of transport of radioactivity to secondary lysosomes (° o) and kinetics of
liberation of iodo-tyrosine from chick muscle cultures (• •) after brief exposure of muscle to
'"I-a-bungarotoxin. Data obtained from electron microscope autoradiographs and from chromato-
graphic analysis of culture medium.
DEPARTMENT OF EMBRYOLOGY
83
Effect of Muscle Activity on
Receptor Turnover
D. J. Card and D. M. Fambrough
Recent experiments have shown that
muscle activity is important in regulat-
ing extrajunctional chemosensitivity of
skeletal muscles. Acetylcholine sen-
sitivity of muscle increases following
denervation and decreases upon rein-
nervation. Electrical stimulation of
denervated or cultured embryonic
muscle can lead to a decline in ex-
trajunctional ACh sensitivity; on the
other hand, an increased sensitivity to
ACh can be produced after blockade of
spontaneous contractions in cultured
cells by tetrodotoxin (a selective in-
hibitor of voltage-sensitive sodium cur-
rent). These changes in ACh sensitiv-
ity reflect changes in the number of ex-
trajunctional ACh receptors. Thus, the
changes should be understandable in
terms of the incorporation of receptors
into the muscle membrane and their
subsequent degradation.
Experimental strategies for measur-
ing incorporation of new ACh receptors
into unstimulated cultured chick
skeletal muscle fibers and the degrada-
tion of receptors thereafter have al-
ready been reported by this laboratory.
We were able to reproduce the observa-
tion of Shainberg and Nelson that tet-
rodotoxin induced a twofold increase in
ACh receptor number within 24 hr of
application to active cell cultures.
However, a quantitative assessment of
the synthesis and degradation of ACh
receptors in active vs. quiescent cul-
tured cells is difficult because it is
nearly impossible to control or to quan-
titate the amount of activity of the cul-
tured cells. After 4 to 5 days in culture,
muscle cells usually begin to develop
spontaneous contractions, but not all
the muscle cells contract. Also, the
time of onset of spontaneous contrac-
tions differs greatly between cultures,
so not all 5-day or even all 11-day cul-
tures are spontaneously active. It has
not been possible to electrically stimu-
late and drive every cell or even most
cells in a culture dish.
This laboratory has previously re-
ported measurements of degradation
rates of extrajunctional receptors from
denervated mouse extensor digitorum
longus (EDL) muscle. Because EDL
muscle is easily activated by electrical
stimulation, it was used to study the
effects of muscular activity on degrada-
tion of receptors. After mid-thigh re-
section of the sciatic nerve, the lower
leg (including the EDL muscle) is de-
nervated. The time course of develop-
ment of extrajunctional ACh receptors
after denervation is tabulated in Table
13. In our studies on receptor degrada-
tion, muscles denervated for at least 5
days were used.
TABLE 13. Accumulation of ACh Receptors
Following Denervation of Mouse EDL
ACh Receptor Sites*
Days After
(molecules x
Denervation
10 9 /mg muscle)
2.0
1
2.0
2
2.1
3
3.5
4
8.8
5
12.5
7
15.6
9
17.6
14
14.1
21
13.6
*Measured as a-bungarotoxin binding sites.
The degradation rate of extrajunc-
tional receptors was inferred from
measurements of the rate at which
[ 125 I]-labeled a-bungarotoxin bound to
these receptors was degraded with con-
comittant release of radiolabeled de-
gradation products from the muscle fi-
bers. We have not yet characterized re-
leased radioactive species, but this has
been done with denervated rat dia-
phragm. Metabolic inhibitors such as
cycloheximide and dinitrophenol and
reduced temperature decreased the
degradation rate, as expected. The per-
fusion chamber used in the degrada-
tion measurements was similar to that
reported last year, but instead of
84
CARNEGIE INSTITUTION
mounting the muscle on a screen it was
tied and stretched between two
platinum electrodes (used for electrical
stimulation) mounted on a Plexiglas
strip. Oxygenated Trowell T-8 medium
was dripped over the muscle and was
continuously aspirated out of the mus-
cle chamber into test tubes; the
radioactivity of the perfusate was then
measured in a gamma counter. The
half-life of ACh receptors in the plasma
membrane averaged 27 hr under these
conditions. After several hours, the
muscle was stimulated at 50 stimuli
per sec for 1 sec every 10 sec for 8 hr.
The degradation rate of receptors de-
creased rapidly to 78% of the value be-
fore stimulation. Thus stimulation of
the denervated EDL muscle increased
the half-life of the receptors in the
plasma membrane to approximately 34
hr.
Degradation rate for extrajunctional
ACh receptors does not appear to
change during the first two weeks of
denervation (Table 14). Also the
stimulation-induced reduction in rate
is independent of the length of the de-
nervation period. Since chemosensitiv-
ity declines during electrical stimula-
tion without any acceleration of degra-
dation, it follows that electrical stimu-
lation must repress the biosynthesis or
incorporation of new receptors into
plasma membranes.
TABLE 14. Degradation Rate of Receptors from
Muscle Denervated 5-13 Days
Denervation Time Receptors Degraded/Hour
(days) (%)
5 2.75
6 2.38
7 2.15
12 2.34
13 2.01
The effects of muscle activity on ACh
receptor biosynthesis and incorpora-
tion into plasma membranes are not
known. We are attempting to perfect a
strategy for the high time-resolution
measurement of incorporation rate of
new receptors into extrajunctional
membranes of denervated adult mus-
cles.
Organization of Muscle Plasma
Membrane in vivo
D. J. Card
Acetylcholine receptors are found
over the entire surface of embryonic
skeletal muscle fibers in vivo. These
sites gradually disappear after inner-
vation, as clusters of acetylcholine re-
ceptors form in the postsynaptic mem-
brane. The junctional receptors may
result from the organization of some of
the extrajunctional receptors into a
much more compact spatial arrange-
ment. In assessing the role of ex-
trajunctional acetylcholine receptors
as components of newly forming
synapses, it is necessary to know
whether junctional receptors appear
before, during, or only as a result of,
formation of functional neuromuscular
synapses. We are studying muscle
morphology and acetylcholine receptor
number and distribution at different
developmental stages of the chick an-
terior latissimus dorsi (ALD) muscles
to learn the time of nerve outgrowth,
synapse formation, appearance of junc-
tional receptors, and disappearance of
extrajunctional receptors.
Muscles were removed from the em-
bryos, fixed in 2% glutaraldehyde, 100
xnM cacodylate buffer, sliced into
1-micron sections and then stained
with toluidine blue. In order to study
receptor binding sites the muscles were
incubated for 1 hr in a solution of
[ 125 I]-a-bungarotoxin (0.5 /xg/ml),
washed several hours and then fixed in
2% glutaraldehyde, 100 mM cacody-
late buffer. Following this treatment,
the total number of acetylcholine re-
ceptors was measured, and single mus-
cle fibers were dissected for au-
toradiography.
Some preliminary observations
based upon light microscopy and au-
toradiography and on the binding of
[ 123 I]-mono-iodo-o;-bungarotoxin follow.
Initially (5-6 days incubation) the pre-
sumptive anterior latissimus dorsi
(ALD) muscle was composed of loosely
organized, undifferentiated cells. How-
ever, after 9 days, muscle cells and
DEPARTMENT OF EMBRYOLOGY
85
nerve bundles were evident. Au-
toradiography of single ALD muscle fi-
bers (6-9 days of incubation) show
radioactive label (ACh receptors) dis-
persed uniformly over the muscle sur-
face. No receptor aggregations were
present. After 13-15 days incubation
the ALD muscle fibers looked mor-
phologically similar to post-hatch mus-
cles, except that fiber size was smaller
and ACh receptors were scattered over
the cell surface. Some areas of each
muscle fiber from 13 to 15 days devel-
opment appeared to have a high recep-
tor density, but these observations
have not been quantified.
At 16 days the ALD had well-defined
clusters of ACh receptors spaced every
120-170 microns along each muscle fi-
ber, with many ACh receptors still
scattered outside the clusters. The
spacing of the dense clusters of ACh re-
ceptors, presumably junctional recep-
tors, agrees favorably with morpholog-
ical, electrophysiological, and cholines-
terase data obtained by others (Gordon,
Perry, Ruffery and Vrbova, Cell Tiss.
Res. 155: 13-25, 1974; Bennett and
Pettigrew, J. Physiol. 241: 515-545,
1974).
Total number [ 125 I]-a-bungarotoxin
binding sites was measured in muscles
from embryonic day 6 to 2 days post-
hatch (Table 15). Most of the [ 125 I] is
present as receptor-[ 125 I]-o;-bungaro-
toxin complexes as judged by sucrose
gradient velocity' sedimentation of de-
tergent extracts. It is not clear yet
when functional synapses are formed.
However, Gordon and her collaborators
and Bennett and Pettigrew have re-
ported morphological nerve contacts
TABLE 15. [ 125 I]-a-Bungarotoxin Binding to
Embryonic Chick ALD Muscles at Different
Developmental Ages
Developmental Age
Muscle Weight (mg)
(days)
cpm/muscle
6
7,000
11
15,300
13
9,800
16
7,700
19
3,200
20
3,000
+ 2
1,300
and acetylcholinesterase staining at
days 9 to 15. The number of ACh recep-
tors per mg muscle is also highest at
that time. As extrajunctional binding
sites disappear, the number of total
ACh receptors/per mg muscle also de-
creases.
Studies on the Mobility of
Extrajunctional ACh Receptors
In Plasma Membrane
R. E. Pagano and D. M. Fambrough
The principal objective of this study,
which is aided by a generous grant
from the Whitehall Foundation, is to
examine the general role of membrane
fluidity in membrane organization and
function. This problem will be tackled
by measuring the lateral (transla-
tional) mobility of acetylcholine recep-
tors in skeletal muscle fibers, and by
studying the effects which a variety of
membrane perturbants (local anesthe-
tics; exogenous phospholipid and
cholesterol treatments) have on this
mobility. Quantitative fluorescence
microscopy will form the basis for de-
termining protein (ACh receptor) mo-
bility. The measurement will be made
by complexing fluorescent-labeled
a-bungarotoxin with ACh receptors on
muscle fibers. A small area of fluores-
cent cell surface will then be bleached
with a short pulse of bright light, and
the rate at which unbleached
molecules from adjacent areas of the
cell surface migrate into the bleached
area will be quantified by measuring
the fluorescence intensity from that
area as a function of time.
During the first year of this study,
some of the necessary background
work for carrying out our experiments
was completed. Fluorescent-labeled
a-bungarotoxin was prepared using
fluorescein isothiocyanate (FITC), and
the preparation was well charac-
terized. It was shown that incubation of
mono-FITC-bungarotoxin, with skele-
tal muscle at 1 /xgm/ml results in
highly fluorescent neuromuscular
junctions as judged by fluorescence
microscopy. The rate of FITC-a-
bungarotoxin binding was also com-
86
CARNEGIE INSTITUTION
pared with that of iodinated
a-bungarotoxin, and found to be about
50-fold slower than the radioactive
species — inconsequential for our exper-
iments. Preliminary experiments
further demonstrated that FITC-a-
bungarotoxin receptor complexes on
skeletal muscle are readily bleachable
and that reversal of this process is a
very slow process.
Preliminary experiments were car-
ried out to test the feasibility of man-
ipulation of cultured muscle cell lipids,
using artificial phospholipid vesicles as
agents for introducing foreign phos-
pholipids into the cell surface. Cells
treated at 2°C with dimyristoyl phos-
phatidyl choline (DMPC) vesicles
showed a significant accumulation of
CMPC in their surface as evidenced by
high resolution electron microscopy
autoradiography. Furthermore, lipid
analyses showed little degradation of
the exogenously supplied lipid. Thus, it
is concluded that significant amounts
of perturbing foreign phospholipids can
be introduced into the cell surface.
Their effect on ACh receptor mobility
remains to be tested.
Preparation of a Fluorescent Derivative
of a-Bungarotoxin
Native a-bungarotoxin binds ex-
tremely tightly to acetylcholine recep-
tors on cultured skeletal muscle cells
and is routinely used as a potent, very
selective, and virtually irreversible in-
hibitor of receptor function. The
radioactive derivative, mono-iodo-a-
bungarotoxin, retains the potency of
the native toxin.
During the past year we have de-
veloped methods for the production of
fluorescent derivatives of a-bungaro-
toxin and have determined the toxicity
of the most useful fluorescent deriva-
tive. In essence the method involves
the interaction of purified a-bungaro-
toxin with fluorescein isothiocyanate
in the cold at pH 9.0 in sodium
carbonate-bicarbonate buffer. An
eight-fold molar excess of FITC is used
and the reaction proceeds slowly, first
forming a mono-FITC-a-bungarotoxin
and later, further derivatized forms.
The unreacted FITC is then separated
from the FITC-a-bungarotoxin by
chromatography on Bio-Gel P-4 and
any unreacted a-bungarotoxin is re-
moved from the FITC-a-bungarotoxin
by ion-exchange chromatography on
CM-Sephadex C-50. It has been found
that the mono- and di-FITC bungaro-
toxin are also separable from each
other in Bio-Gel P-4, the di-FITC
a-bungarotoxin being retarded by the
column. When mono-FITC a-bungaro-
toxin is fractionated on Sephadex C-50,
using a gradient of NaCl in 3.3 raM
phosphate bufferpH 7.2, it is eluted far
ahead of unreacted a-bungarotoxin and
thus quantitatively freed of this con-
tamination.
Properties of Fluorescent
a-Bungarotoxin
The purified mono-FITC a-bungaro-
toxin binds to acetylcholine receptors
with a very high affinity. After incuba-
tion of muscles in a solution containing
about 1 fJLg/m\ FITC a-bungarotoxin,
followed by rinsing to remove unbound
toxin, the acetylcholine receptors at
the neuromuscular junctions are com-
plexed with FITC-cc-bungarotoxin and
appear as vividly fluorescent struc-
tures when examined by fluorescence
microscopy. This interaction is essen-
tially irreversible, and these fluores-
cent structures maintain their fluores-
cence for more than 6 months at 10°C
in a dilute buffer solution.
The relative rates of interaction of
FITC-a-bungarotoxin and of radioac-
tive iodinated a-bungarotoxin have
been compared by mixing these two
toxin derivatives in different concen-
tration ratios and then measuring the
amount of iodinated a-bungarotoxin
that bound to cultured skeletal muscle.
Such competition curves have been
carried out on three preparations of
FITC-a-bungarotoxin, and all suggest
that the speed of binding is reduced
50-fold for the fluorescent derivative.
Thus, higher concentrations are re-
DEPARTMENT OF EMBRYOLOGY
87
quired to saturate the acetylcholine re-
ceptors in a short incubation period.
Even so, 10" 7 M FITC-a-bungarotoxin
will saturate the receptors in cultures
of skeletal muscle in about 20 minutes.
Use of FITC-a-Bungarotoxin as
a Probe in Studying Membrane
Structure
Following our original proposal we
have developed the FITC-a-bungaro-
toxin as an easily bleachable fluores-
cent tag for cholinergic receptors. Pre-
liminary experiments using micro-
scopic equipment at the Leitz and Zeiss
showrooms in New Jersey and New
York and qualitative experiments done
with our yet incomplete fluorescence
microscopy set-up at Carnegie demon-
strate that fluorescent FITC-a-bun-
garotoxin-receptor complexes on
skeletal muscle fibers are readily
bleached and that the bleaching is
mainly irreversible or only very slowly
reversible. Also, stable fluorescence
measurements can be made when ap-
propriate filters and light chopping are
employed. Thus the main prerequisite
for a fluorescent probe to use in execut-
ing our proposed experiments on the
mobility of receptors in plasma mem-
branes has been met.
Turnover of Plasma Membrane
Proteins
K. Tepperman
The model for the turnover of the
ACh receptor of muscle cells raises the
question of the relationship of the re-
ceptor to other plasma membrane pro-
teins. The study of ACh receptor turn-
over is aided by the availability of the
specific probe a-bungarotoxin. Since
other membrane proteins are not so
easily identified, new approaches to
studying turnover of populations of
proteins are being devised. One ap-
proach is to covalently label surface
proteins by enzymatic iodination and
to follow loss of radioactivity from spe-
cific proteins. Another approach, de-
veloped by Robert Schimke, is to allow
the cells to incorporate different
radioactive isotopes of leucine on two
successive days and to look at the ratio
of the different isotopes for each of the
membrane proteins. We are using both
of these approaches to study the turn-
over of plasma membrane proteins of
cultured muscle cells.
In Year Book 74, p. 78, we reported
preliminary observations on enzymatic
iodination of muscle cells. At that time
the total material on iodinated tissue
culture dishes was extracted in SDS
and run on polyacrylamide gels, result-
ing in approximately 15 radioactive
bands. It is difficult to determine which
of these proteins are actually part of
the membrane and which proteins may
be less intimately associated with the
surface. In the past year, several
methods for dissociating loosely bound
proteins from membranes were ex-
plored. One currently in use is treat-
ment of cells with low ionic strength
medium, followed by homogenization
and centrifugation to obtain a mem-
brane pellet. When this pellet is used
for SDS gels, the resulting pattern of
iodinated polypeptide chains is much
simpler than when total material is
used. The number of radioactive bands
is reduced to approximately seven, and
one of the major bands in the original
preparation, one with a molecular
weight of about 230,000 daltons, is
among those which are lost, suggesting
that this protein is not an integral
membrane protein.
Preliminary results using surface
iodination for turnover studies have
suggested that most of the proteins
that are accessible to iodination are
degraded at the same rate. In these ex-
periments, sets of cultures are iodi-
nated, then incubated at 37 °C for dif-
ferent periods of time. After extraction,
equivalent amounts of radioactivity
from different chase periods are loaded
onto gels. Densitometer traces of au-
toradiographs of the gels show that dif-
ferent bands have essentially the same
relative amounts of radioactivity. Ex-
periments are now in progress in which
[ 13, I]-bungarotoxin is bound to cells
88
CARNEGIE INSTITUTION
that have been treated with lac-
toperoxidase and [ 125 I]. This is done to
provide an internal measure of the rate
of ACh receptor turnover with which to
compare the loss of [ 125 I]-label from
other membrane proteins of the cells.
Initial results are compatible with the
idea that turnover of the iodine-labeled
proteins occurs at the same rate as
turnover of the ACh receptor.
In the iodination experiment, several
points have been raised which require
further clarification. Although pub-
lished reports on lactoperoxidase iodi-
nation of cells generally indicate little
labeling of lipids, when muscle cells
are iodinated, approximately 35% of
the radioactivity in a membrane frac-
tion is chloroform methanol extracta-
ble, indicating that lipids have been
labeled. Thin-layer chromatograms
have been done which show that the
label is present in a variety of lipid
fractions. Another result of the iodina-
tion experiments is that during the
course of incubation following iodina-
tion, the radioactivity lost from the
cells can be fractionated into three
components by column chromatog-
raphy on Bio-Gel P-2. One fraction
probably represents nonspecifically as-
sociated iodide, which comes off the
cells and the tissue culture dishes dur-
ing the incubation. Another fraction is
iodotyrosine, which probably results
from internalization and degradation
of labeled proteins. A third fraction,
excluded by the P-2 columns, may rep-
resent proteins or other material that
is lost from the cells.
Another approach to studying turn-
over of populations of proteins is the
double isotope method of Schimke. In
his reports using this method, he gen-
erally finds a correlation between
turnover rates of proteins and their
molecular weights. Preliminary results
using this method with cultured mus-
cle cells show that the correlation for
the major polypeptides of a crude
membrane preparation holds. How-
ever, experiments are in progress
which will study this question on
plasma membranes specifically. Also,
the effects of various treatments de-
signed to remove all but integral mem-
brane proteins will be studied to de-
termine whether the molecular weight
correlation with turnover rate still
holds. Such a correlation is contraindi-
cated by the iodination results.
Studies on Neuronal Plasma
Membranes
S. Carbonetto
Much of our work on the plasma
membrane during differentiation of
skeletal muscle has relied upon the use
of a-bungarotoxin and its radioactive
derivatives that bind selectively to the
acetylcholine receptor molecule in
muscle cells. Similar information
about formation of neuronal plasma
membranes has been unobtainable due
to lack of a suitable ligand. This report
deals with our initial studies on a
neuronal preparation that apparently
binds a-bungarotoxin with high affin-
ity.
Sympathetic ganglionic chains have
been dissected out of 11 to 12 day old
chick embryos, dissociated in 0.25%
trypsin, washed and plated onto
collagen-coated culture dishes in
Eagles MEM plus Nerve Growth Fac-
tor (2 units/ml; Burroughs Wellcome).
Within 24 hr neurons become attached
to the surface of the plate and begin to
send out processes. Intracellular elec-
trophysiological recordings from these
cells after three days indicate that
neurons have resting membrane poten-
tials of 30 mV, and can be stimulated to
generate action potentials (Fig. 54).
Furthermore, when cultured chick
sympathetic neurons are incubated
with [ 125 I ]-a-bungarotoxin and
analyzed by light microscope au-
toradiography, grains appear over the
neuron soma and down to the tip of the
processes but are absent over non-
neuronal cells in the culture (Fig. 54).
Washing the culture for 28 hr at 14 °C
after [ 125 I]-a:-bungarotoxin binding
only slightly reduced the density of
grains over neurons. In one experiment
DEPARTMENT OF EMBRYOLOGY
89
B
100 y
B
I
+i&
"***&,
Fig. 54. (A) Intracellular recording from a chick sympathetic neuron in culture. The voltage record,
V, shows an action potential evoked by a stimulating pulse, S, applied to the surface of the neuron by
an extracellular microelectrode. The uppermost trace is at mV. The calibration is 50 mV; 20 msec.
(B) Autoradiograph of cultured chick sympathetic neurons incubated with 125 I-a-bungarotoxin. When
viewed with phase optics the neuron and several nonneuronal cells are visible in the field (B,). With
bright field illumination (B 2 ) only exposed emulsion grains are visible and these cover the neuronal
cell body and its neurites. There is no localization of grains over any of the nonneuronal cells.
10~ 3 M d-tubocurarine prevented the
association of radioactivity with the
cells.
These preliminary data suggest that
chick sympathetic neurons in culture
exhibit properties of well-differen-
tiated neurons and also may specif-
ically bind a-bungarotoxin. Experi-
ments are in progress to determine the
association and dissociation kinetics of
90
CARNEGIE INSTITUTION
bungarotoxin binding to neurons and
also to localize the appearance of the
grains by EM autoradiography. We are
testing whether these neurons contain
functional acetylcholine receptors by
electrophysiologically monitoring their
response to iontophoretically applied
acetylcholine and testing if such re-
sponse can be blocked by application of
bungarotoxin.
SIGNALING IN THE NERVOUS SYSTEM: SYNAPSES OF SPECIFIC
NEURONS IN THE LEECH
K. J. Muller and S. T. Carbonetto
with the technical assistance of B. Thomas
The role that a neuron assumes in
the nervous system is dependent upon
the connections it makes with other
neurons or with peripheral targets.
Therefore, an essential step both in de-
termining the forces that act initially
to organize the nervous system and in
characterizing the mechanisms by
which a neuron handles, or integrates,
information is to ascertain the rela-
tionship between a neuron's function
and the arrangement of its synapses. A
related problem in neurobiology has
been to determine whether a neuron's
shape and other properties are intrin-
sic or whether its contacts or associa-
tions with other neurons are essential
for certain aspects of its growth and
maintenance.
The nervous system of the leech
Hirudo medicinalis has been particu-
larly favorable for study of the physio-
logical role and functioning of specific
neurons, largely because its organiza-
tion is relatively simple. Neurons in
this segmented worm are grouped 350
together to form essentially identical
ganglia that are linked in a chain by
axon bundles; each ganglion mediates
the reflexes and controls the behavior
of its segment. It has been possible to
monitor electrical activity from within
single sensory and motor neurons with
microelectrodes and to correlate the
physiological activity of functionally
identified neurons with their charac-
teristic shape and position within each
ganglion. By recording from pairs of
neurons one can trace synaptic connec-
tions, which have been found to be sur-
prisingly uniform and in that sense
predictable between specific neurons.
Last year we described studies under
way that provided a morphological de-
scription of sensory and motor neurons
and of the distribution of their
synapses within the ganglion (Year
Book 74, p. 85). These investigations,
using the enzyme horseradish
peroxidase (HRP) as an intracellular
marker for both light and electron mic-
roscopy, were to lay the groundwork for
our determining any structural basis
for physiological changes that we had
previously produced with selective le-
sions of the leech nervous system (J.
Physiol. (Lond) 242, 289, 1974), as well
as ft)*- charting the course of regenerat-
ing neurons. An additional feature of
HRP as an intracellular marker is that
although it diffuses extensively within
an injected neuron, it does not pass be-
tween electrically coupled neurons, as
do smaller markers used by others.
Neuron terminals that make electrical
contacts are therefore sharply demar-
cated even within an intricate
meshwork of neuron processes.
Chemical Synapses in the Leech
K. J. Muller and B. Thomas
Structural Details of the Synapse
A good picture of the normal, fully
developed synapse is requisite for the
study of the distribution and formation
of synapses. Because the chemical
synaptic transmission that one can
record in the leech central nervous sys-
DEPARTMENT OF EMBRYOLOGY
91
tern resembles that in other inverte-
brates and vertebrates, it is not sur-
prising that in the electron microscope
customary synaptic features — the pre-
sence of presynaptic vesicles, mem-
brane densities and a widened, uniform
intercellular cleft — serve to delineate
the synapse. In the leech, synapses are
made almost exclusively in the center
of the ganglion in the region called the
neuropil rather than on cell bodies that
form layer or cortex around the gan-
glion. Sections through the neuropil
manifest a variety of synaptic profiles,
but careful analysis reveals that most
result from different planes of section
through the same structural configura-
tion (Fig. 55). Typically, agranular ves-
icles (—50 nm diameter) are clustered
near a ridge of presynaptic density that
apposes a pair of conjoint postsynaptic
profiles across a cleft about 30 nm
wide. Larger (—100 nm) dense-core
vesicles and mitochondria are also
found in the presynaptic terminal.
Identified mechanosensory neurons
that have been injected with the HRP
for an intracellular marker have
branched processes that exhibit similar
features (Fig. 56). This indicates, in ac-
cord with their physiology, that these
neurons are presynaptic. The same
processes are also postsynaptic to other
neurons. In contrast with the sensory
neurons, the large longitudinal (L) and
annulus erector (AE) motoneurons that
act to shorten the animal and to erect
its annular rings into ridges, respec-
tively, are seen in the electron micro-
scope to be solely postsynaptic within
the neuropil. Physiologically these are
neurons upon which the mechanosen-
sory neurons synapse. One class of ap-
parently presynaptic terminals (Fig.
59, white arrow) has been found that
does not readily fit the pattern of ter-
minals that contain agranular vesicles:
These terminals, which contain small
granular vesicles (—50 nm diameter)
and extremely dense, eccentric cored
large vesicles (— 100 nm), are generally
found to be cells that stain with the dye
Neutral Red and are widely regarded
to contain serotonin as a transmitter.
What are in fact the membrane spe-
cializations at synapses made by such
neurons remains to be determined.
The Arrangement of Synapses on
Sensory Neurons
One remarkable pattern in the or-
ganization of the leech nervous system
is that synapses of mechanosensory
neurons are clustered upon fingerlike
swellings or enlargements. Their line-
aments are sufficiently similar within
a modality, e.g., touch, pressure or
stimulus, but different between them
to permit a reliable distinction to be
made between sensory modalities
solely on the basis of the shapes and
distribution of synaptic terminals
viewed in the light microscope. A pos-
sibility raised by the differences in the
shapes of synapse clusters is that, as
with certain vertebrate motor neurons,
the morphologies of terminals are in-
trinsic to the presynaptic neurons and
are not dependent upon the type of
postsynaptic target. This will be dis-
cussed in the next section.
From a functional standpoint, the
separation of clusters from one another
and the difference in synaptic inputs to
them suggest that clusters may some-
times act autonomously. But this
means a single neuron that synapses
onto two neurons in two spatially sepa-
rate regions could produce differential
postsynaptic effects. We have reported
such differential actions {J. Physiol.
(Lond) 238, 357, 1974), and studies are
under way to determine if indepen-
dence of clusters from one another can
account for these and other integrative
phenomena.
Sprouting of Sensory Neuron
Terminals
The paired connectives that link ad-
jacent ganglia to form the leech nerve
cord are axonal pathways that are vir-
tually free of synapses. Certainly the
mechanosensory neurons, which syn-
apse in their own neuropil and in adja-
cent ganglia, run directly between
92
CARNEGIE INSTITUTION
Fig. 55. Top: General scheme of synaptic specializations in the leech, based on serial sections. The
presynaptic terminal contains numerous 50-nm agranular vesicles (only those clustered next to the
plasma membrane are shown) and is separated from postsynaptic processes by extracellular space
widened to about 30 nm. Lateral to a dense band along the inner surface of the presynaptic mem-
brane, an occasional vesicle fuses with the membrane. Bottom: A band in such a terminal, cut in cross
section, is indicated by an arrow.
ganglia without branching and synaps-
ing until they pass the connective. It
was recently reported that crushing a
leech connective can cause synapses to
form within the connective at the
crush. This presents us with a system
that promises to make it possible to
examine the formation de novo of
DEPARTMENT OF EMBRYOLOGY
93
Fig. 56. Sites of synapses of the pressure, P, touch, T, and nociceptive, N, sensory cells culled from
serial electron micrographs of regions like those indicated by arrowhead in Fig. 57 a and b. Three-
dimensional reconstructions of injected neurons (upper left) were projected (upper right) and stippled
to indicate approximate distributions of synaptic vesicles within the irregular fingers that arise from
secondary processes. Points of contact onto pairs of postsynaptic processes are indicated by large dots;
at the triangles the sensory neuron is postsynaptic.
synapses made by functionally iden-
tified neurons whose developed syn-
apses have now been well described.
We have found that in animals
whose connectives have been severed
or crushed between a pair of ganglia
several months previously and not
permitted to regenerate, the crushed
portion of sensory axons has in some
cases sprouted a profusion of branches
at the site of the cut or crush (Fig. 57).
The branching unfortunately occurs
within a mass of scar tissue that be-
comes dense after staining for specific
neuron processes, so comparisons were
made with a cleaner, in vitro system
developed by others. In culture,
neurons also sprout at the crush, but
scar tissue is not present to obscure the
details of branching (Fig. 57). We have
seen no qualitative differences in vivo
and in vitro with the light microscope.
This system provides an ideal chance to
examine the growth of new processes
and the requirements for synapse for-
mation. In the light microscope, we can
now say that sensory neurons form
swellings that look like presynaptic
terminals along and at the ends of new
branches, and that such branches form
94
CARNEGIE INSTITUTION
not only at a crush or cut between gan-
glia, but even at free cut end. In oper-
ated animals a comparison has not yet
been made between modalities, but the
shapes of enlarged fingers along new
branches of touch-sensitive (T) cells do
resemble normal T-cell presynaptic
terminals. In culture a clear distinction
cannot be made between the new T-cell
arbors and the new branches of
pressure-sensitive (P) neurons. The
complex and distinctive cluster of en-
larged fingers of the P cell within the
neuropil (see Fig. 57) has not developed
at the crush for periods up to 3
weeks — a time sufficient for regenera-
tion across the cut. It is now necessary
to examine the sequence of growth and
to determine if some of the regenera-
tion we have seen might involve a fu-
sion of a growing axon with the old,
distal stump. Whether it is a loss of its
target, direct damage to its axon, or the
presence of new targets that fosters a
neuron's sprouting is testable with a
variety of lesions, particularly in cul-
ture.
It remains unclear whether some en-
Fig. 57. Experimentally induced sprouting of the sensory neuron axon in the connective. A crush or
cut was made between two ganglia as shown at left. Sensory neurons were injected at their cell bodies
with HRP, stained with diaminobenzidine, and their processes traced from whole mount using camera
lucida. (a) and (b): Normal secondary branches of pressure, P, and touch, T, neurons within the
ganglion neuropil. Arrowheads indicate where synapses are typically clustered, for comparison with
swellings along new branches in the connective (c) to (f). (c): Sprouting where the connective was cut
100 days previously, (d) and (e): Crush between ganglia maintained in culture for 24 days,P, and 13
days, T. (f): Crush at end of connective rather than between ganglia (24 days in culture), (g): Regener-
ation across cut at 24 days in culture. Note the fewer sprouts. T andP placed at ends nearest cell body;
dashed line, edge of connective and lesions.
DEPARTMENT OF EMBRYOLOGY
95
largements at early stages might rep-
resent growth cones. One likely pros-
pect is that such enlargements will be-
come synaptic terminals in a new
neuropil formed at the site of the crush
as reported in the connective of another
species of leech. We have not yet ob-
served structures that can be clearly
identified as synapses, but the presence
of large dense-core and small agranu-
lar vesicles at ends of branches of iden-
tified sensory neurons suggests that
synapse formation may be in its early
stages. In this case, the structures that
make synapses take shape before
synapses are made and would appear to
dictate the association of synapses that
are peculiar to sensory cells.
An Electrical Synapse in the
Leech
K. J. Muller and B. Thomas
Both types of neuronal synapses —
chemical and electrical — are readily
detected physiologically with mic-
roelectrodes in the leech. Yet in leech
ganglia electron microscopy has failed
to provide us with a clear example be-
tween neurons of gap junctions, the
putative sites of electrical transmis-
sion. The reason for our failure re-
mains obscure, but it is likely that the
actual extent of synaptic membrane is
limited and that the separation be-
tween membranes is, as at other inver-
tebrate gap junctions, wider than at
tl
25p
D
30 \l
Fig. 58. Rohde fibers extend to meet in the connective. (A) Rohde neuron injected in the cell body in
one ganglion reaches halfway to the next. The region of overlap with the uninjected Rohde cell in the
lower ganglion is included in (B). When both are injected and a cross section taken through such a
region, axons of the two cells (A, and A 2 ) can be resolved. (C) light micrograph; and (D) corresponding
electron micrograph of boxed region in (C). The axons run in the small, unpaired nerve (Faivre's
nerve).
96
CARNEGIE INSTITUTION
gap junctions in vertebrates. It was
therefore unexpected when an electri-
cal junction was discovered between
unpaired neurons in neighboring gan-
glia, neurons that were reportedly syn-
cytial. The junction across which dye
molecules as large as Procion yellow
(MW about 500) can readily pass but
not horseradish peroxidase (MW about
40,000) occurs between "giant" Rohde
fibers in the middle of the unpaired
connective (Fig. 58). The electrical
coupling between adjacent Rohde
neurons is so strong that no junction
was thought to exist between them, yet
it is only at this mid-connective region
that the two neurons meet and
synapse. Visible in the electron micro-
scope is the extensive arborization of
the two cells into fine, connecting pro-
cesses, rather than an extensive con-
tact along a septum (Fig. 59). The de-
tails of the contact between the
neurons is still under intensive exami-
nation, but the region of overlap along
the connective is nearly 200)U, and the
extent of the many contacts is nearly as
great. Also peculiar to this region of
apposition are numerous terminals
(Fig. 59, white arrow) containing small
and large granular vesicles of the sort
that have been shown by others to be
inhibitory upon leech muscle. Such an
inhibitory action upon the fine Rohde
fiber collaterals could have pronounced
and perhaps directional effects upon
electrical transmission between Rohde
fibers. This possibility is currently
under physiological and anatomical
investigation using selective stimula-
tion and marking techniques.
Interruption of the Rohde Neuron
The discovery that Rohde neurons
contact each other across an electrical
junction that is physically separate
from the neuropil and is permeable
enough for some large molecules to
pass has presented us with an unusual
opportunity to examine the interde-
pendence of the two neurons.
For example, by crushing or severing
a connective near one ganglion, the
distal portion of one Rohde axon is iso-
lated from the soma. Can it then be
maintained through its synapse? Or
does the interruption stimulate the
Fig. 59. The electrical synapses between Rohde filters in the connective. The principle axons (1 and
2) branch profusely as they terminate (left). The cells contact each other principally along their
branches (right) at areas of narrowed extracellular space (black arrow) 3-4 nm wide as at gap
junctions in some other invertebrates. Terminals containing small granular vesicles (white arrow)
are seen in close association with the branches. Only neuron 1 has been injected with HRP, but axon 2
is recognizable by its size and location.
DEPARTMENT OF EMBRYOLOGY
97
other intact Rohde fiber to grow beyond
the synapse until it can reestablish
functional contact with the adjacent
Rohde neuron? Our results are pre-
liminary, but where we have been able
to examine this problem, we have
found in an animal operated upon 79
days previously that the undamaged
neuron does grow beyond the synapse.
The neighboring Rohde neuron, its
axon severed and unable to reestablish
contact, has undergone a massive and
unusual arborization within the gan-
glion while eliminating much of its
large axon (Fig. 60). Whether the new
processes are making functional con-
tacts is not clear, but it is noteworthy
that the level of spontaneous synaptic
and impulse activity that can be re-
corded from such cells is elevated, while
the basic features of the nerve impulse
are preserved.
A study of the role that this synapse
can play in the structure and
metabolism of the interacting cells is in
progress using ganglia in culture.
Here, by selectively destroying one
neuron and thereby eliminating prin-
cipal target tissue, the relative impor-
tance of intrinsic and extrinsic stimuli
to growth and synapse maintenance in
the nervous system may be weighed.
Normal
Crushed
Connective
79 days
crush
Fig. 60. Plasticity of the Rohde interneuron. The connective between ganglia was crushed in an
animal 79 days prior to injection of the neuron, identified by the location of its soma and the size and
shape of its action potential. The principal axon has disappeared beyond the open triangles, even
within the anterior connective that was untouched, and new sprouts have appeared (black triangles)
that reach toward the connective but only slightly beyond the ganglion.
THE COLLECTION OF HUMAN EMBRYOS
Ronan O'Rahilly and Ernest Gardner
The Carnegie Embryological Collec-
tion was opened formally at the Uni-
versity of California in Davis on 21
November 1975. Drs. Ronan O'Rahilly
98
CARNEGIE INSTITUTION
and Ernest Gardner were in atten-
dance, and speakers included Drs. J. D.
Ebert, Elizabeth M. Ramsey and Ar-
thur T. Hertig. Dr. Ramsey presented a
book on the Carnegie monkey colony,
and Dr. Hertig later donated the
binocular dissecting microscope he had
used in his searches for early human
embryos. Dr. George W. Corner was
unable to be present. The University
was represented by Deans Allen G.
Marr, C. John Tupper, and William R.
Pritchard. Carnegie President Philip
H. Abelson visited the laboratories on
the following day.
All inquiries concerning the collec-
tion, as well as requests for permission
for publication, should be addressed to
Professor R. O'Rahilly, Carnegie Labo-
ratories of Embryology, University of
California, Davis, California 95616.
Dandy-Walker and Arnold-Chiari mal-
formations," "The prenatal develop-
ment of the human eye," and "The tim-
ing and sequence of events in the de-
velopment of the limbs in the human
embryo" were published. An account of
"The nerve supply and conducting sys-
tem of the human heart at the end of
the embryonic period proper" is due to
appear in the Journal of Anatomy.
Studies of the developing anterior fal-
cate artery are being undertaken by
Dr. F. Muller.
The Bluntschli Collection
Developmental Stages in Human
Embryos
Work continues on the revision of
stages 10 to 23 (R. O'Rahilly), on the
tabulation of the timing and sequence
of developmental events in staged
human embryos (R. O'Rahilly), and on
the establishment of a computer
catalogue (Alexander Barry).
Development of the Nervous System
Work continues on the develop-
mental neurobiology of primates (E.
Gardner and associates). A chapter on
"The developmental anatomy and his-
tology of the human central nervous
system" is awaiting publication in the
Handbook of Clinical Neurology, edited
by Vinken and Bruyn. Articles on "The
Renewed interest has been shown in
the comparative collection of the late
Professor Hans Bluntschli of Bern. It
was established on expeditions to
Amazonia in 1912 and to Madagascar
in 1930, and was acquired by the Car-
negie Institution in 1950. The
Bluntschli collection comprises insec-
tivores (tenrecs), prosimians (lemurs),
and platyrrhines (New World mon-
keys), as shown in Table 16. During the
past year, the Bluntschli microscope
slides were studied by Dr. F. Muller
(Basel) and by Prof. F. Strauss (Bern).
Professor Strauss had worked with
Professor Bluntschli in Switzerland
and is donating further material rele-
vant to the collection. Professor David
B. Meyer has been studying the com-
parative development of the eye.
Theses (for the M.Sc. degree) on "Em-
bryonic limb development in the
rhesus monkey (Macaca mulattaV by
G. Stone and on "Staging of the early
embryonic brain in the baboon (Papio
sp.) and rhesus monkey (Macaca
mulatto)" by R. Davignon have been
approved.
DEPARTMENT OF EMBRYOLOGY
TABLE 16. Chief Insectivores and Primates in the Carnegie Collection
99
BLUNTSCHLI COLLECTION
Insectivora
Centetes (Common tenrec)
Hemicentetes (Banded tenrec)
Ericulus (Hedgehog tenrec)
Prosimii
Microcebus (Mouse lemur)
Avahi laniger (Woolly lemur)
Platyrrhinae
Saguinus fuscicollis (Hapale; Mystax
bluntschlii) (Tamarin)
Aotes trivirgatus (Nyctipithecus vociferans)
(Owl-faced or night monkey)
Alouatta senculus (Mycetes auratus)
(Red howler-monkey)
Saimiri sciurea (Chrysothrix nigrivittata)
(Squirrel monkey)
Cebus apella (C. macrocephalus) (Hooded
or brown capuchin)
Cebus albifrons (C. gracilis) (White-fronted
capuchin)
Lagothrix lagotricha (L. infumatus) (Brown
woolly monkey)
HARTMANN COLLECTION
Catarrhinae
Macaca mulatta (Macacus rhesus)
(Rhesus macaque)
MALL COLLECTION
Catarrhinae
Homo sapiens
Chiarelli's (1972) Decimal Key
Suborder
Family
Genus
Species
1.
4.
1.
1
1.
5.
2.
1
2.
1.
4.
3
2.
3.
1.
1
2.
3.
6.
3
2.
3.
7.
1
2.
3.
8.
1
2.
3.
8.
3
2.
3.
11.
1
STAFF ACTIVITIES
International conferences and sym-
posia in which members of the De-
partment participated during the year
included three in Scotland ("The
Synapse," in St. Andrews, sponsored by
the Scottish Electrophysiological Soci-
ety; "Genetic Engineering," in Glas-
gow, the Second Tenovus Symposium
of the Nucleotide Group, Chemical and
Biochemical Societies; and the annual
symposium of the British Society of
Developmental Biologists, also in
Glasgow). The annual symposium of
the Japanese Society of Developmental
Biologists at Osaka also proved to be
attractive. Others attended symposia
held before and during the meeting of
the Federation of European Biochemi-
cal Societies in Paris, one on "Or-
ganelle Biogenesis" and another on
"Structure of the Eukaryotic Genome."
Similar topics, "Genetic Function of
Mitochondrial DNA" and "The Organi-
zation and Expression of Chromo-
somes," were treated in Bari, Italy, and
in Berlin, respectively. Also deserving
mention was the Institut de la Vie's
Conference on "Teratogenesis" in Mar-
tinique, and the Symposium on
"Biochemistry of Cell Surfaces" held at
the Fogarty Center of the National In-
stitutes of Health, Bethesda.
100
CARNEGIE INSTITUTION
Among the other conferences engag-
ing the interest of members of the
group, the following should be men-
tioned: the Cold Spring Harbor Sym-
posium on "The Synapse"; the annual
Symposium of the Society of General
Physiologists, on the subject
"Biogenesis and Turnover of Mem-
branes"; a related symposium on
"Membrane Receptors" at Duke Uni-
versity; a meeting sponsored by the
Neurosciences Research Program (Bos-
ton) on "Depolarization-Release Cou-
pling Systems in Neurons"; Gordon Re-
search Conferences on "Developmental
Biology" and "Molecular Pharmacol-
ogy"; and the ICN-UCLA Symposium
on "Molecular Mechanisms in the Con-
trol of Gene Expression."
Lectures were presented on a
number of campuses, including Albert
Einstein College of Medicine, Califor-
nia Institute of Technology, Case
Western Reserve University, Columbia
University, Cornell Medical College,
DePaul University, Harvard Univer-
sity, Hershey Medical Center, Middle-
bury College, Rockefeller University,
University College, London; and the
Universities of Alberta, Arizona, and
Bern; California at Davis, Irvine, Los
Angeles, San Diego, and San Fran-
cisco; Chicago, Cincinnati, Connec-
ticut, Iowa, Kyoto, Maryland, Michi-
gan, Pennsylvania, Rochester, Tokyo,
Utah, and Virginia; Vanderbilt Uni-
versity, Washington University, and
Yale University.
Special presentations included The
John M. Prather Lectures in Biology at
Harvard, the Mitsubishi Lecture in
Tokyo, and a Lowell Lecture in Boston.
Members of the group also spoke at
hospitals and research centers, among
them the Radiation Effects Research
Foundation in Hiroshima, Japan; the
Beatson Institute for Cancer Research,
Glasgow; the Roche Institute for
Molecular Biology; Oak Ridge As-
sociated Universities; Sloan-Kettering
Institute; National Institutes of
Health; and the Muscular Dystrophy
Research Laboratories at Newcastle-
upon-Tyne.
Members of the Department took
part in meetings of learned societies,
including, in addition to those already
mentioned, the American Association
for the Advancement of Science,
American Association of Anatomists,
American Society of Biological
Chemists, American Society for Cell
Biology, American Philosophical Soci-
ety, Biophysical Society, Federation of
American Societies for Experimental
Biology, National Academy of
Sciences, Society for Developmental
Biology, Society of Neurochemistry,
Society for Neuroscience, and Tissue
Culture Association.
Advisory and consultative services
included Editor-in-Chief of Develop-
mental Biology and membership on the
editorial boards of Anales del Desar-
rollo, Cell, Excerpta Medica (section on
Human Developmental Biology), Cur-
rent Topics in Developmental Biology,
Oceanus, and Quarterly Review of Biol-
ogy.
Members of the staff also acted in
these capacities: Chairman, Assembly
of Life Sciences, National Academy of
Sciences-National Research Council;
Member of the Board of Governing
Trustees and the Board of Scientific
Overseers, Jackson Laboratory; Trus-
tee, President, and Director, Marine
Biological Laboratory; Member of Cor-
poration, Woods Hole Oceanographic
Institution; Trustee of International
Medical Congress, Ltd., and member of
its program committee planning the
Fifth International Congress on Birth
Defects.
Other posts occupied by members of
the Department include the following:
in the American Institute of Biological
Sciences, member, Council of Past
Presidents and member, Finance
Committee; in the Institut de la Vie,
Chairman, World Committee on the
Formative Weeks of Human Life; in
the National Academy of Sciences,
member, Committee on Energy; in the
National Institutes of Health, member,
Board of Scientific Counselors, Na-
tional Institute of Child Health and
DEPARTMENT OF EMBRYOLOGY
101
Human Development, member, Geron-
tology Research Center Resources Ad-
visory Committee, member, Physiolog-
ical Chemistry Study Section, and
member, Cell Biology Study Section;
and in the Society for Developmental
Biology, Past-President.
Members of the Department served
on several visiting committees, includ-
ing those for the Departments of Biol-
ogy at Harvard University, Mas-
sachusetts Institute of Technology, and
Princeton University; the Department
of Molecular Biology and Biochemistry
at Harvard; and the External Advisory
Committee of the University of
Chicago's Cancer Research Center.
Special service was rendered as co-
Chairman of the "Cluster on Develop-
mental Biology" of the President's
Panel on Biomedical Research.
Staff members taught courses at the
Marine Biological Laboratory as well
as at Johns Hopkins (in the Depart-
ments of Biology, Biophysics, Pediat-
rics and Physiological Chemistry).
Seminars
The roster of speakers at the semi-
nars organized by the Department to
serve all those working in develop-
mental biology in the region included:
Aftab Ahmed, National Naval Medical
Center; S. T. Carbonetto, University of
North Carolina; James Darnell, Rocke-
feller University; Sarah Elgin, Har-
vard University; Walter Gilbert, Har-
vard University; Robert Grainger,
Yale University; Alan F. Horwitz,
University of Pennsylvania; Karl
Illmensee, Institute for Cancer Re-
search, Philadelphia; Lionel Jaffe,
Purdue University; Stuart Kauffman,
University of Pennsylvania; M. S.
Letinsky, University of California at
Los Angeles; D. K. Morest, Harvard
Medical School; Josh Sanes, Harvard
Medical School; Susumu Tonegawa,
Basel Institute for Immunology; Her-
man Vandenburgh, University of
Pennsylvania; David Ward, Yale Uni-
versity; and K-W Yau, Stanford Uni-
versity.
102
CARNEGIE INSTITUTION
BIBLIOGRAPHY
Adler, W. H.,see Ozato, K.
Albright, J. F., J. W. Deitchman, S. A. Has-
sel, and K. Ozato, Differential antibody
production by adherent and nonadherent
spleen cells transferred to irradiated and
cyclophosphamide- treated recipient
mice. J. Reticuloendothelial Soc., 17,
195-209, 1975.
Biroc, S. L., and R. H. Reeder, Iodination of
Xenopus laevis histone F2al in chroma-
tin. Biochemistry, 15, 1440-1448, 1976.
Bokhon'ko, A., and R. H. Reeder, The sub-
unit structure of mouse satellite chroma-
tin. Biochem. Biophys. Res. Coram., 70,
146-152, 1976.
Brown, D. D., Genome organization in
higher organisms, Federation Proceed-
ings, 35, 11-12, 1976.
Brown, D. D., see also Brown, R. D., Carroll,
D., and Thomas, C.
Brown, R. D., and D. D. Brown, The nuc-
leotide sequence adjoining the 3' end of
the genes coding for oocyte-type 5S
ribosomal RNA in Xenopus. J. Mol. Biol.,
102, 1-14, 1976.
Carroll, D., and D. D. Brown, Adjacent re-
peating units of Xenopus laevis 5S DNA
can be heterogeneous in length, Cell, 7,
477-486, 1976.
Carroll, D., and D. D. Brown, Repeating
units of Xenopus laevis oocyte-type 5S
DNA are heterogeneous in length, Cell,
7, 467-475, 1976.
Dawid, I. B., see Leister, D. E., and Wel-
lauer, P. K.
Deitchman, J. W., see Albright, J. F.
Devreotes, P. N., and D. M. Fambrough,
Synthesis of acetylcholine receptors by
cultured chick myotubes and denervated
mouse extensor digitorum longus mus-
cles. Proc. Nat. Acad. Sci., 73, 161-164,
1976.
Devreotes, P. N., and D. M. Fambrough,
Turnover of acetylcholine receptors in
skeletal muscle, Cold Spring Harbor
Symp. Quant. Biol, 40, 237-251, 1976.
Ebert, J. D., "Katsuma and Dan and the
Quiet Revolution," in Development,
Growth and Differentiation, Vol. 18, pp.
i-ii, 1976.
Ebert, J. D., see also Ozato, K.
Fambrough, D. M., see Devreotes, P. N.,
and Ritchie, A. K.
Gardner, E., and R. O'Rahilly, Neural crest,
limb development, and thalidomide em-
bryopathy, Lancet, 1, 635-637, 1976.
Gardner, E., R. O'Rahilly, and D. Prolo, The
Dandy-Walker and Arnold-Chiari mal-
formations, Arch. Neurol., 32, 393-407,
1975.
Gardner, E., see also O'Rahilly, R.
Hassel, S. A., see Albright, J. F.
Huang, L., and R. E. Pagano, Interaction of
phospholipid vesicles with cultured
mammalian cells. I. Characteristics of
uptake. J. Cell Bio/., 67, 38-48, 1975.
Huang, L., see also Pagano, R. E.
Leister, D. E., and I. B. Dawid, Mitochond-
rial ribosomal proteins in Xenopus
laevislX. mulleri interspecific hybrids, J.
Mol. Biol, 96, 119-123, 1975.
O'Rahilly, R., The prenatal development of
the human eye, Exp. Eye Res., 21, 93-
112, 1975.
O'Rahilly, R., and E. Gardner, The timing
and sequence of events in the develop-
ment of the limbs in the human embryo,
Anat. Embryo!.., 148, 1-23, 1975.
O'Rahilly, R., see also Gardner, E.
Overton, C. G., see Weinberg, E. S.
Ozato, K., Review of J. Klein, "Biology of
the mouse histocompatibility-2 complex.
Principles of immunogenetics applied to
a single system." Science, 190, 978, 1975.
Ozato, K., W. H. Adler, and J. D. Ebert,
Synergism of bacterial lipopolysac-
charides and concanavalin A in the acti-
vation of thymic lymphocytes, Cellular
Immunology, 17, 532-541, 1975.
Ozato, K., J. D. Ebert, and W. H. Adler,
Pretreatment of thymocytes by
phytohemagglutinin inhibits the binding
of :3 H-Concanavalin A, J. Immunol., 115,
334-339, 1975.
Ozato, K., and J. D. Ebert, Concanavalin A
potentiates syngeneic response in
murine lymphocytes, J. Exp. Med., 143,
1-14, 1976.
Ozato, K., J. D. Ebert, and W. H. Adler, The
differentiation of suppressor cell popula-
tions as revealed by studies of the effect
of mitogens on the mixed lymphocyte
reaction and on the generation of
cytotoxic lymphocytes, Cell Immunol.,
22, 32.3-333, 1976.
Ozato, K.,see also Albright, J. F.
Pagano, R. E., and L. Huang, Interaction of
phospholipid vesicles with cultured
mammalian cells. II. Studies of mecha-
nism, J. Cell Biol., 67, 49-60,1975.
Pagano, R. E., see also Huang, L.
Prolo, D., see Gardner E.
DEPARTMENT OF EMBRYOLOGY
103
Reeder, R. H., see Biroc, S. L., Bokhon'ko,
A., and Weinberg, E. S.
Ritchie, A. K., and D. M. Fambrough, Ionic
properties of the acetylcholine receptor
in cultured rat myotubes. J. Gen.
Physiol., 65, 751-767, 1975.
Ritchie, A. K., and D. M. Fambrough, Elec-
trophysiological properties of the mem-
brane and acetylcholine receptor in de-
veloping rat and chick myotubes, J. Gen.
Physiol., 66, 327-355, 1975.
Shutt, R. H., see Weinberg, E. S.
Suzuki, Y., Fibroin messenger RNA and its
genes, Adv. Biophys., 8, 83-114, 1975.
Thomas, C. and D. D. Brown, Localization
of the genes for silk fibroin in silk gland
cells of Bombyx mori, Developmental
Biology, 49, 89-100, 1976.
Weinberg, E. S., C. G. Overton, R. H. Shutt,
and R. H. Reeder, Histone gene ar-
rangement in the sea urchin, Stron-
gylocentrotus purpuratus, Proc. Nat.
Acad. Sci., U.S.A. 72, 4815-4819, 1975.
Wellauer, P. K., and I. B. Dawid, Structure
and processing of ribosomal RNA: A
comparative electron microscopic study
in three animals, Brookhaven Symposia
in Biology, 26, 214-223, 1974.
PERSONNEL
Year Ended June 30, 1976
(including those whose services began or ended during the year)
Research Staff
Donald D. Brown, Biochemistry,
Director-designate
Igor B. Dawid, Biochemistry
James D. Ebert, Director
Douglas M. Fambrough, Biochemistry
Kenneth J. Muller, Neurobiology
Richard E. Pagano, Biophysics
Ronald H. Reeder, Biochemistry
Yoshiaki Suzuki, Biochemistry
Research Associates (extramural)
Bent G. Boving, Detroit, Michigan
Robert L. DeHaan, Atlanta, Georgia
Ernest Gardner, Davis, California
Arthur T. Hertig, Boston, Massachusetts
Irwin R. Konigsberg, Charlottesville,
Virginia
Ronan O'Rahilly, Davis, California
Elizabeth M. Ramsey, Washington, D.C.
Postdoctoral Fellows and
Grant-Supported Associates
Peter Botchan, U.S. Public Health Service
Grant (Reeder)
Diana Card, Fellow of the U.S. Public
Health Service; Fellow of Muscular
Dystrophy Associations of America, Inc.
S. T. Carbonetto, Fellow of Carnegie
Institution of Washington
Jeffrey Doering, Fellow of the U.S. Public
Health Service
Scott Emmons, Fellow of the National
Cystic Fibrosis Research Foundation
Nina Fedoroff, Fellow of the U.S. Public
Health Service and Carnegie Institution
of Washington
Paul Geshelin, Fellow of the U.S. Public
Health Service
Elizabeth Godwin, U.S. Public Health
Service Grant (Dawid)
Leaf Huang, Fellow of Carnegie Institution
of Washington and the U.S. Public
Health Service
Yasumi Ohshima, Fellow of Carnegie
Institution of Washington
Keiko Ozato, Joint Fellow of Carnegie
Institution of Washington and Marine
Biological Laboratory
Aileen K. Ritchie, Fellow of Muscular
Dystrophy Associations of America, Inc.
Alex Sandra, U.S. Public Health Service
Grant (Pagano)
Masatoshi Takeichi, Fellow of Carnegie
Institution of Washington
Katherine Tepperman, Fellow of the U.S.
Public Health Service
Harvey Wahn, U.S. Public Health Service
Grant (Reeder)
104
CARNEGIE INSTITUTION
Peter K. Wellauer, Fellow of the National
Cystic Fibrosis Research Foundation
Students
Peter Devreotes, Graduate, Johns Hopkins
University
John Gardner, Graduate, Johns Hopkins
University
Carol Kaushagen, Graduate, Johns
Hopkins University
Jose Ramirez, Graduate, Johns Hopkins
University
Visiting Investigators and
Extramural Collaborators
W. H. Adler, Baltimore, Maryland
George Brownlee, Cambridge, England
John W. Chase, Boston, Massachusetts
Hayden G. Coon, Bethesda, Maryland
Michael Edidin, Baltimore, Maryland
J. W. S. Harris, London, England
Uel J. McMahan, Boston, Massachusetts
R. M. Ruysschaert, Brussels, Belgium
Christian Thomas, Brussels, Belgium
Witold Wozniak, Poznan, Poland
Clerical and Technical Staff
Elaine S. Asch, Senior Technician
Judith M. Bernstein, Secretary
James H. Blackwell, Custodian
Leo Blackwell, Recorder
Paul Blackwell, Custodian (part-time)
Julian L. Certain, Librarian (part-time)
Huei-Wen Chen, Technician
Jeffrey Ciemny, Recorder
William J. Cleary, Recorder
William H. Duncan, Senior Technician
Ernestine V. Flemmings, Laboratory
Helper
Kate Francis, Secretary
Paul Giza, Technician
Richard D. Grill, Photographer
Virginia S. Hicks, Laboratory Helper
Mary E. Hogan, Technician (part-time)
John E. Jones, Custodian
Eddie D. Jordan, Senior Technician
Carrie Kahn, Technician (part-time)
Catherine R. Lane, Librarian (part-time)
Alice H. Mabin, Laboratory Helper
Thomas F. Malooly, Business Manager
Rudolph Matthews, Recorder
Judy Miller, Recorder
Thomas F. Miller, Custodian
Ann N. Murphy, Secretary
Joyce I. Patterson, Laboratory Helper
John Pazdernik, Jr., Building Engineer
Betty Lou Phebus, Bookkeeper-Technician
Martha L. Rebbert, Senior Technician
Adrienne Robinette, Secretary
Bessie H. Smith, Laboratory Helper
Delores V. Somerville, Senior Technician
John L. Wiser, Machinist
Department
of Terrestrial Magnetism
Washington, District of Columbia
George W. Wetherill
Director
Carnegie Institution of Washington Year Book 75, 1975-1976
Contents
Introduction 109
Galaxies and Stars 118
Evidence for motion of the Galaxy 118
A search for extragalactic extinction 119
The statistics of galaxy alignments 121
Kinematic studies of NGC 3351, 3115, and 1275 122
An illustration of galaxy redshifts 127
The irregular galaxy WLM and its surroundings 130
The nucleus of M31 133
Star counts in globular clusters 134
Neutral hydrogen supernova remnant study 136
Type I supernovae: Interpretation of
optical spectra 139
The 21-cm receiver for the Instituto Argentino
de Radioastronomia 143
820-MHz survey of the Southern sky 144
Image tube systems 145
Activities of the Instituto Argentino de
Radioastronomia 147
The Earth and the Solar System 149
Gas-solid interactions in the solar nebula 149
Interplanetary bodies and solar system history 151
The role of large impacting bodies in the
formation of the earth and moon 152
Late (4.0 b.y.) bombardment of the moon
and terrestrial planets 157
The earth-crossing Apollo objects 164
Particle tracks and micro-impact features
in grains from gas-rich meteorites 170
Cosmic-ray research 176
Rb-Sr mantle isochrons 176
Part I: Continental regions 177
Part II: Oceanic regions 191
Geochemistry of late Cenozoic volcanic rocks from
the Nevado de Toluca area, Mexico 207
Preliminary Rb/Sr data on the minimum age of the
central Andean Precambrian basement complex 213
Volcanoes, Q , and seismicity in western
South America 216
Determination of the lithosphere-asthenosphere
boundary using converted waves 223
Shallow seismicity in subduction zones 229
A model for laterally heterogeneous velocity
structure at the base of the mantle 233
Single site phase velocity determinations 240
Maximum entropy spectral analysis 243
Atomic Physics and Properties of Materials 249
Diffusion in silicate melts and glasses 249
Diffusion measurements using fast deuterons for
in situ production of radioactive tracers 259
Observation of x radiation characteristic of
a united atom 263
Bibliography 268
Personnel 270
INTRODUCTION
It has become a rather tiresome
cliche to contrast the universal genius
or "Renaissance man" of former years
with today's specialist. Concentration
of one's work into limited areas of
knowledge is often blamed for the fre-
quent, inability of scientists to grasp
adequately the relationships between
their work and that of others, and to
achieve the unifying syntheses which
are the goal of intellectual endeavor.
Worse yet, in the political and enter-
tainment world simple mention of the
titles of specialized research projects
results in serious scholars becoming
sources of public amusement or objects
of public ridicule.
In this context it is sometimes forgot-
ten that there is a unifying countercur-
rent upon which even the most spe-
cialized scientists are borne. The main-
stream of twentieth century physics
and chemistry has been the develop-
ment of a universal theory of matter,
applicable to all times and all places. In
addition to blurring the boundary be-
tween physics and chemistry almost to
the point where the difference is unrec-
ognizable, this development has been
of profound importance to all of natural
science. Astronomy, geology, and biol-
ogy have lost their taxonomic and de-
scriptive flavor.
For a time those in the vanguard of
this development felt it necessary to
identify their activity (often in a
somewhat superior tone) as geophysics
or geochemistry rather than geology,
or as astrophysics rather than as-
tronomy. (Even between these so-
called interdisciplinary fields there is a
certain hierarchy. I once annoyed a
famous solar spectroscopist by intro-
ducing him to a class of chemistry stu-
dents as an astrochemist; upon gaining
the floor he firmly reestablished his
identity as an astrophysicist.) Al-
though these distinctions are still
made, they are less useful, and possibly
as part of the current vogue for nostal-
gia, there are now many scientists
primarily concerned with ther-
modynamics of multiphase systems
who proudly describe themselves as
geologists, and experts in nuclear reac-
tions who would object to being called
anything but astronomers. In many
universities, if you really want to learn
thermodynamics you will attend lec-
tures in the geology department, and if
you really want to learn fluid mechan-
ics you will go to meteorology or
oceanography; and the genuine physics
and chemistry departments find them-
selves only one among many and not
always the best. Thus the battle is
over; physics and chemistry are now
the cornerstone of all science. The
earth is now treated as a chemical and
mechanical system; stars and galaxies
are to be understood only through
knowledge of plasmas, nuclear reac-
tions, and molecular chemistry.
However, even though physicists and
chemists have reproduced themselves
within geology and astronomy depart-
ments to the point where their progeny
are now dominant, it is also apparent
that the unity in the basic approach to
science which has now been achieved
has not been as fully exploited as it
could be, and that many of the old bar-
riers remain, even though they are
more often surmounted today than in
the past.
The reasons for the continued exis-
tence of these obstacles are manifold.
To a significant extent they are a con-
sequence of practical problems bas-
ically of a nonscientific nature. In a
university, lecture courses have to be
taught, and students have to be accred-
ited as bona fide geologists, astrono-
mers, or meteorologists if they wish to
find employment. The department
needs to be sure it gets at least its fair
share of university positions, funds,
and space. The need to deal with these
109
110
CARNEGIE INSTITUTION
problems can unite the faculty of a de-
partment more effectively than their
mutual scientific interests, and it de-
flects their attention from what might
become mutually stimulating scientific
interaction with those in other de-
partments. In another area, the sheer
size of the federal funding structure re-
quires agencies such as the National
Science Foundation to be com-
partmented into "disciplines," even
though this word has lost most of the
significance it formerly possessed. Pro-
gram directors and the associated peer
reviewers can impose severe penalties
on those scientists who stray too far
from what may be regarded as the most
fundamental and important subjects
for research even though they may
simply be the most fashionable at the
moment within a limited group of sci-
entists.
We have been given the opportunity
to be remarkably free of these essen-
tially nonscientific pressures to iden-
tify with a particular professional
group. Our more informal mode of in-
struction avoids the problem of re-
peatedly "covering" a prescribed body
of knowledge. Our endowed research
support and our freedom from govern-
ment influence is much greater than
that available to our colleagues
elsewhere, and historically there has
been within the Department a tradi-
tion of extending our research beyond
the point which a reviewer might well
regard as the limits of our competence.
Freedom from duties associated with a
professorial position and the obligation
to serve on institutional and federal
committees also can provide us with
the time and tranquility necessary to
look inward to the Department, to un-
derstand and value the work of one
another, and to contemplate in greater
depth the meaning of our work and its
relationship to that of others. With this
unusual opportunity comes the respon-
sibility to break away from artificial
boundaries associated with our profes-
sional origins and to build a research
community dedicated to viewing the
earth, solar system, stars, and galaxies
as a continuous and unified field of sci-
entific investigation. Although at
greater risk, workers elsewhere have
taken steps to achieve this. The princi-
pal barrier to our success in this effort
should be only our finite abilities.
The reader of this report can judge
the extent to which we have accepted
this responsibility. Some changes from
previous reports can be seen. There are
no longer organized reports of "sec-
tions" of astrophysics, geophysics, and
atomic physics. The report has been
subdivided into units only in order to
make it easier to read. The boundaries
between these units are intentionally
arbitrary. New work has begun in an
area of science transitional between
our previous work in astronomy and
earth science. This is the study of the
solar system and its history, as re-
vealed by the most primitive material
available, the meteorites, and by
theoretical investigations ultimately
based upon observations of the earth,
the moon, and the planets. Of course,
all of this could be viewed as a facade; a
perceptive reader will note that to a
large extent the same people are work-
ing on the same problems as before. A
cynical reader could conclude that all
of the foregoing is a pretense and that
things will always be the same. That is
not our intention. Only time will tell.
Current Research in the
Department
There is a natural hierarchy in the
organization of the universe. On the
largest scale there are the clusters of
galaxies, galaxies, and their con-
stituent stars and interstellar mate-
rial. Within the stars are formed the
elements that are the matter from
which new stars and planetary systems
are in turn formed. The formation of
new stars and planetary systems is
controlled by the physical and chemical
structure of the interstellar medium. It
is the structure of the interstellar
medium that defines the initial and
boundary conditions for the formation
of planetary bodies such as the earth
DEPARTMENT OF TERRESTRIAL MAGNETISM
111
and moon. These conditions largely de-
termine the nature and magnitude of
the processes which have occurred dur-
ing subsequent earth history, and
which have resulted in the present
chemical and physical state of our
planet. Underlying and unifying all
are the fundamental properties of mat-
tter as learned from atomic and nuclear
physics.
At all levels in this hierarchy we are
confronted with mysteries. Although
the earth has been extensively studied,
one need go only a few tens of kilome-
ters beneath its surface to arrive at the
frontier area of current research. The
formational solar system has tended to
be treated as a magical wonderland
where one qualitative speculation is as
good as another. Although much prog-
ress has been made in understanding
stellar evolution during the past sev-
eral decades, there does not exist at the
present time any usable quantitative
theory for understanding either the
beginning or the end of a star's life. In
the area of galactic evolution we are at
best emerging from a descriptive stage
where masses, velocity fields, and
chemical abundances are measured for
various types of objects including
"peculiar" ones in the expectation that
before too long these data will form the
basis for a more general understanding
of galaxies. The strong coupling be-
tween the nature and history of
galaxies, stars, and planets results in
this imperfection of understanding in
one area affecting our ability to under-
stand another. Some sort of simultane-
ous attack on the unknown is needed.
We are obviously too small to under-
take this mighty task. However, in
working simultaneously at several of
these frontiers along with other scien-
tists throughout the world we are in an
excellent position to contribute to the
I solving of these problems and to the
understanding of their relationship to
Galaxies and Stars
During the past year investigations
of the universe at the galactic scale
have made use of the all-sky sample of
Scl galaxies which was established by
Ford, Rubin, and Thonnard in their
study of the anisotropy of the Hubble
expansion. This work has now been
completed, with the conclusion that the
observed anisotropy has its origin in
motion of both our Galaxy and the
Local Group of galaxies, relative to the
more distant sample galaxies, at a ve-
locity of about 450 km s" 1 . The conflict
between this result and the reported
isotropy of the 3°K relic black body
radiation remains as a problem to be
resolved theoretically or observation-
ally.
Now that optical spectra and radio
observations exist for such an all-sky
sample, they can be used for the inves-
tigation of other fundamental prob-
lems. Rubin and Thonnard have in this
way explored the possibility that be-
tween the members of the group of
nearby galaxies there is an excess of
extragalactic matter, analogous to the
interstellar material within a galaxy.
Such material would be expected to lie
within the plane defined by these
nearby galaxies, and should lead to ex-
cess extinction in the direction of this
plane, compared with the perpendicu-
lar direction. No significant excess ex-
tinction is found. It is concluded that
the density of this intergalactic mate-
rial is similar to the average density of
the universe, and that no significant
portion of the missing matter required
to cause the universe to be gravitation-
ally closed was found in the form of this
"intergalactic medium."
In past exploration of the nature of
the various types of galaxies, principal
attention has been directed toward ob-
servation of spiral galaxies, their veloc-
ity fields and masses being determined
by observation of bright emission lines
of the gas that is abundant in these
galaxies. The velocity fields of the un-
derlying stars that form the bulk of the
mass of galaxies have been relatively
little studied in the spirals and in the
equally important gas-free elliptical
galaxies. Ford, Peterson, and Rubin
report on the use of absorption lines
112
CARNEGIE INSTITUTION
originating in the stars. Velocity fields
and mass and angular momentum
models have been obtained for two ob-
jects, a barred spiral galaxy and an
early-type galaxy. It is found that the
rotation over the radial extent of the
bar is that of a solid body, but whether
the stars are permanently associated
with the bar or the bar is a density-
wave phenomenon passing through the
stellar distribution cannot be deter-
mined from the data. Overall the
dynamics of the barred galaxy are very
similar to those of a normal galaxy. For
the first time, high resolution spectra
of an elliptical galaxy are available to
provide a sound observational basis for
constructing a realistic dynamical
model. In addition, investigations of a
peculiar colliding pair of galaxies dis-
covered by Minkowski 20 years ago
show that the galaxies are still ap-
proaching one another, and the evi-
dence for an earlier explosive event in
one of the galaxies cannot be explained
by the presence of the other galaxy.
The all-sky sample of galaxies has
also been used by the same workers to
illustrate the good general agreement
of both optical and 21-cm meas-
urements of redshifts when measured
on the same galaxies.
Ford, ^Peterson, and Rubin also re-
port on the more detailed structure of
the nearby Andromeda galaxy, M31.
They find evidence that the angular
momentum vector of the nucleus may
be tilted with respect to the plane of the
disk of the galaxy. If this is the case,
this will be an important fact to be in-
cluded in attempts to understand the
general principles of the dynamics of
galaxies. Other interpretations are
possible. However, future observations
should permit these alternatives to be
tested.
Thonnard reports 21-cm observa-
tions of neutral hydrogen in the irregu-
lar galaxy WLM, which lies in proxim-
ity to the Magellanic Stream. He finds
that this galaxy, although containing
relatively few stars, contains almost as
much neutral hydrogen as the Large
Magellanic Cloud, and that this hydro-
gen is located within WLM rather than
being a portion of the adjacent Magel-
lanic Stream.
Turner has carried out a Monte
Carlo study of the statistics of galaxy
alignments in order to understand the
significance of reported observations of
such alignments. He finds that the
probability of such alignment by
chance is surprisingly high, and that a
fair number of such alignments should
occur even in the absence of a genetic
or dynamic association between the
aligned galaxies.
Peterson has continued his work on
globular clusters, and reports new star
counts and comparisons with dynami-
cal models for 27 clusters. Globular
clusters are perhaps the oldest type of
object available for investigation and
may predate the galaxy with which
they are associated. Their structure
may therefore still bear a record of the
earliest events of galactic history.
These data provide constraints which
must be satisfied by theories of galactic
evolution.
Going to more recent events in the
history of the Galaxy, new results on
supernovae are described. An under-
standing of these explosion events is
essential to understanding the termi-
nal stages of stellar evolution, element
formation, x-ray sources, pulsars,
neutron stars, cosmic ray sources, and
possibly the initiation of the formation
of new stars. Consequently there has
been considerable recent interest in the
development of supernova models.
Such models require observational
data against which they may be tested.
Assousa reports calculations which
may explain why in his previous work
neutral hydrogen was found in only a
few of the supernova remnants studied.
Use of remnant expansion velocities
from other workers permits calculation
of the mass of the objects studied and
shows good agreement between the
predicted and the observed neutral hy-
drogen mass.
Assousa, Ford, Peterson, and Rubin
describe new spectra of extragalactic
type I supernovae and their compari-
DEPARTMENT OF TERRESTRIAL MAGNETISM
113
son with generated synthetic spectra.
They conclude that these spectra ap-
pear to be dominated by transitions in
ionized iron, which could possibly be
produced by radiative excitation dur-
ing the explosive event.
In the area of instrumentation, much
effort of both the scientific and techni-
cal staff (Ecklund, Little, Thonnard,
and Turner) is going into the construc-
tion of a modernized 21-cm receiver for
use on the 30-meter radio telescope at
the Instituto Argentino de Radioas-
tronomia. It is expected that this will
permit much higher quality observa-
tions of Southern sky high velocity
clouds and spiral galaxies than has
hitherto been possible. Digital equip-
ment for an 820-MHz survey of the
Southern sky is also under construc-
tion.
Ford continues his work with image
tube systems, in addition to his col-
laborative use of the present "Carnegie
Image Tubes" in the current investiga-
tions described earlier in this report.
Two new spectrographic image tube
systems have been constructed in our
shop. In addition, facilities are under
development for analysis and evalua-
tion of the new generation of digital
imaging devices as they become avail-
able.
A brief report is also given of the ac-
tivities of the Instituto Argentino de
Radioastronomia, with which we at
DTM have been closely associated
since its beginning. In spite of severe
economic difficulties, a significant pro-
gram of instrument development and
observation has continued. Of particu-
lar significance has been the near com-
pletion of a low velocity neutral hydro-
gen survey which reveals the large-
scale structure of this material over
most of the Southern sky.
The Earth and the Solar System
At the interface between astronomi-
cal science and earth science lies inves-
tigation of the formation of the solar
system. The processes are adjuncts of
star formation. Many of the facts
against which theories can be tested
are based on measurements of rocks,
particularly meteorites.
In this area of science there is a se-
vere need for the development of
theories sufficiently quantitative to be
testable and communicable. There
needs to be developed a community of
scientists who have enough respect for
the work of one another to seriously at-
tempt to understand its details,
criticize what they believe to be wrong,
and build upon what is of value. Unfor-
tunately such a community does not
exist at present; rather too frequently,
investigations are pursued in splendid
isolation from one another, and while
there is much disagreement, there is
too little effort to understand the origin
of these differences.
Weidenschilling has made signifi-
cant progress in development of a
quantitative model of the dynamics of
particles of various sizes moving in a
gaseous disk of the type generally be-
lieved to have been present during the
early stages of planet formation. This
builds upon and extends earlier work of
Whipple. He shows that there is a
strong dependence on particle size
which may influence the rate at which
the initial ^ 1-km bodies are formed
and possibly also the chemical compo-
sition of these bodies.
Somewhat later in the formation of
the solar system these earlier conden-
sations must somehow or other have
coagulated to form planets. There has
developed in the past five years or so a
common belief that the time scale for
this accumulation must have been very
short, on the order of 10 3 to 10 4 years.
There exists at present no quantitative
dynamical theory which predicts such
a rapid time scale. Perhaps the princi-
pal reason for widespread acceptance of
this short time scale has been the evi-
dence that at least much of the moon
was heated above its melting point es-
sentially at the time of its formation,
leaving insufficient time for the slow
buildup of temperature by the decay of
long-lived radioactive elements and
therefore requiring the trapping of
114
CARNEGIE INSTITUTION
gravitational energy associated with
rapid accretion. In 1969, V. S. Saf-
ronov, working at the Institute of the
Physics of the Earth in Moscow, pub-
lished a theory of early terrestrial
heating in which a calculated distribu-
tion of large impacting bodies during
accretion "buried" a sufficient quantity
of their interplanetary kinetic energy
beneath the accreting surface of the
earth to heat much of the interior to
near its melting temperature. We were
fortunate in having the opportunity of
hearing Safronov explain this work
further at a recent DTM seminar.
When applied to the moon, however,
this theory still fails to provide the
necessary initial temperature, primar-
ily because in the case of the earth the
interplanetary kinetic energy is aug-
mented by acceleration during the fall
of the body into the earth's gravita-
tional field. This effect will be much
less for the smaller moon.
During the past year a variant of
Safronov's theory has been developed
(Wetherill) using techniques previ-
ously employed by the author for the
study of the evolution of meteorite or-
bits. This leads to a considerably
greater fraction of the accreting bodies
being large (>200 km in radius) and of
high velocity (~8 km/s). Both of these
result in much greater burial of kinetic
energy during the formation of the
moon, and thereby provide a mecha-
nism for obtaining initial high lunar
temperature even if the time scale is as
large as that given by quantitative
dynamical theories (~10 8 yr).
Application of the same and similar
techniques to two other solar system
problems are reported. These are the
causes of the late heavy bombardment
(4.0 b.y. ago) which produced the
crater-saturated surface of the moon
(and probably other planets as well) of
that age and the role played by present
earth-crossing interplanetary bodies
(Apollo objects) as sources of chondritic
meteorites.
Extension of realistic theoretical in-
vestigations into early solar system
history, both here and elsewhere, may
be expected to clarify considerably our
thinking with regard to these prob-
lems, but taken alone, are not likely to
be able to distinguish between alterna-
tive hypotheses. Fortunately they do
not have to be taken alone, as there are
available samples of meteoritic rocks
which in various senses have preserved
relic information concerning the ear-
liest history of the solar system and
quite likely of some presolar events as
well. In addition, at least some meteor-
ites have been affected by more recent
events prior to their impacting the
earth, and these must be disentangled
somehow from the primordial effects in
order to be well understood. A promis-
ing opening into this area was found by
Brownlee and Rajan several years ago
in the discovery of submicron impact
craters on the surfaces of meteorite
grains. When combined with the oc-
currence of solar flare tracks in the
same grains, this leads to an analogy
between the surface of the meteorite
parent body and the lunar regolith
(soil), for which such phenomena have
been extensively studied. Rajan and
his collaborator Poupeau of C.N.R.S.,
France, report a continuation of this
work, directed toward an understand-
ing of similarities and differences be-
tween the lunar regolith and the in-
ferred regolith surfaces of the parent
bodies of meteorites of various kinds.
Lunar studies show that the proba-
bility of a grain found on the surface
being cratered increases as the regolith
matures, as a consequence of being
continually impacted, turned over, and
mixed. These new results show the
meteoritic data lie on the extrapolation
of this lunar trend and correspond to
very immature regoliths. This correla-
tion suggests a common dynamical
mechanism that is operative on all re-
goliths. On the other hand, glassy
splashes are absent from meteorite
grains, although they are quite abun-
dant even on immature lunar soils, an
effect which is not yet understood.
Other extraterrestrial research re-
ported here concerns measurement of
reversals of the sun's poloidal magnetic
DEPARTMENT OF TERRESTRIAL MAGNETISM
115
field as inferred from ionization cham-
bers operated continuously by the De-
partment since 1937 (Forbush). This
time scale of almost three decades is
long compared with that of changing
emphases in the field of high energy
physics with which cosmic-ray re-
search was associated in 1937, but it
represents less than two cycles in the
variations of the sun's poloidal field.
The success of these measurements is a
tribute to the value of making observa-
tions on the time scale peculiar to the
phenomena being studied rather than
on the basis of changing fashion.
Since the beginning much of the re-
search of this Department has been
strongly oriented toward global prob-
lems of the earth and its history. This
continues to be the case. Currently
major emphasis is being placed upon
using geochemical and geophysical
techniques to learn as much as possible
concerning the nature of the earth's
mantle.
During the past year, much of the
time of those working in isotope
geochemistry (Brooks, James, and
Hofmann, in collaboration with Hart,
now at M.I.T.) has been devoted to tak-
ing stock of the isotopic data developed
here and elsewhere in recent years,
and attempting to understand its im-
plications regarding the earth's mantle
beneath the oceans and the continents.
A compilation of all the data of suffi-
cient quality shows that in both these
regions the initial isotopic composition
of the strontium mantle-derived rocks
is not uniform. The initial 87 Sr/ 8H Sr
ratios of rocks from both oceanic is-
lands and continental areas are usu-
ally distinctly higher than the much
more uniform values found in mid-
ocean ridge basalt. Such elevated Sr
isotope ratios have commonly been in-
terpreted as indicating either a crustal
origin of the rocks or at least crustal
contamination. This interpretation is
particularly untenable for oceanic
rocks; arguments are presented
against this being the case for most of
these continental rocks as well.
A key point in these arguments is
that in a given region the data indicate
not only higher initial Sr isotope ratios
but a correlation between the 87 Sr/ 86 Sr
ratio and the Rb/Sr ratio. Thus a
"pseudoisochron" is obtained which re-
sembles the more familiar whole-rock
isochrons which are used for age de-
terminations and are based on the pre-
sent 87 Sr/ 86 Sr ratio of the rocks rather
than their initial ratio. This correla-
tion is interpreted as indicating that
the present rocks possess not only a
"memory" of some mean Rb/Sr ratio of
their source as indicated by their ini-
tial Sr isotope ratio but that some in-
formation concerning the time scale for
the variation of this Rb/Sr ratio in the
mantle source has also been preserved.
Thus "pseudoisochrons" are in fact
mantle isochrons and may possess real
time significance, even though
obscured by scatter in the data result-
ing from the superposition of later
events. More specific models are pro-
posed for the chemical development of
these mantle source regions. The seis-
mic evidence for the great thickness of
the continental lithosphere and the
possible complete absence of continen-
tal asthenosphere is incorporated into
these models, and it is hypothesized
that the continental mantle isochrons
are the result of more or less closed sys-
tem evolution of an ancient heteroge-
neous continental lithosphere, with a
typically higher Rb/Sr ratio than the
more depleted asthenosphere sources of
ocean ridge basalts. In the oceanic is-
lands, there is not thick lithosphere of
this kind, and it is proposed that the
deeper mesosphere plays an analogous
role in the formation of these rocks.
In addition to this synthesis of
isotopic data obtained during the past
several years, some new results are
also reported. Whitford and his col-
laborator Bloomfield in the Institute of
Geological Sciences in London are
working on orogenic volcanic rocks
from Mexico which are situated on old,
typically continental crust adjacent to
a convergent plate boundary. The rocks
are compared with typical island arc
116
CARNEGIE INSTITUTION
volcanics from the Sunda arc of In-
donesia. The Mexican rocks appear to
be geochemically distinct from those of
the island arc. They are richer in Si0 2
and Na 2 and are characterized by
higher atomic Mg/Mg+Fe 2+ ratios. Al-
though this work is still in progress,
the isotopic data now obtained together
with the high Mg/Mg + Fe 2+ ratios
suggest a mantle origin for the Mexi-
can rocks.
James has obtained a minimum
Rb/Sr whole-rock age of 1012 ± 52 m.y.
for rocks from the central Andean
basement complex near Arequipa,
Peru. This is the oldest age measure-
ment reported from this basement
complex and suggests a similar or older
age for the mantle lithosphere underly-
ing this region, which is believed to be
the source of those rocks exhibiting the
mantle isochrons discussed earlier in
this introduction.
These geochemical results are re-
lated to geophysical data in the same
area. Sacks, Snoke, and Linde sum-
marize their work on the distribution
of volcanism, mantle anelasticity, and
seismic activity in South America.
They find that in southern Peru and
Chile there is an anelastic zone (low Q)
between the subducting oceanic plate
and the >300 km thick continental
lithosphere. This is correlated with
seismic activity in the downgoing plate
and presence of volcanoes, indicating
partial melting of the mantle in this
region. Farther to the north in Peru,
the low-Q zone is absent, the subduct-
ing slab has low seismicity, and vol-
canoes are absent. It appears that the
development of a well-defined seismic
plane in the subducting slab of oceanic
lithosphere near its interface with the
rocks above requires these overlying
rocks to be hot and partially molten.
Both geophysical and geochemical
data indicate more and more that old
continental areas are underlain by
lithospheres hundreds of kilometers
thick, and thus despite the occurrence
of continental drift on moving plates,
continents are not an extremely super-
ficial feature of the earth. This gives
rise to the question of whether a lower
boundary to the continental litho-
sphere exists at all, and if so, if it can
be detected. Conventional techniques,
using surface waves, have no resolu-
tion below about 200 km depth. Sacks
and Snoke describe seismic arrivals in
Peru, northern Chile, and Argentina
which they have been able to model
only as converted phases arising from a
nearby horizontal transition zone at a
depth of 400 km. Furthermore, these
data indicate the seismic shear velocity
is lower below the interface, which
leads to the interpretation that a boun-
dary between the thick continental
lithosphere and an underlying as-
thenosphere has been found.
Sacks, Rodriguez, Snoke, and Linde
report results from the long-term col-
laboration between DTM and the Uni-
versity of San Agustin in Arequipa,
Peru. These data show that there is an
earthquake-free shallow (<50 km
thick) region on the continental side of
the trench extending for about 80 km.
This same feature is found elsewhere,
e.g., in the Aleutians. The explanation
offered for this phenomenon is that al-
though the rocks in these regions are of
normal strength they are subjected to a
superposition of stresses (a compres-
sion resulting from the relative plate
motion and a tension caused by the
curvature of the plates at the trench)
which leads to the region being nearly
stress free.
These results in observational seis-
mology have been supported by several
interesting theoretical studies. In 1974
Sacks reported anomalies in the
amplitudes of seismic phases that have
traversed the earth's core, and he
showed that these are correlated with
certain regions on the core-mantle
boundary. Thus in some unclear way
these results were maps of "geogra-
phic" regions on the core-mantle inter-
face. Snoke and Sacks now propose a
velocity model that fits these data. In
this model the anomalous regions con-
sist of cells ~ 150 km in dimension
DEPARTMENT OF TERRESTRIAL MAGNETISM
117
which multiply scatter the seismic sig-
nals traversing them along near-
grazing ray paths. The size and veloc-
ity variations required are of a mag-
nitude compatible with their being
small convection cells. These cells are
believed to occur in clusters in those
areas of the lowermost mantle corre-
sponding to the anomalous regions
found earlier.
Sacks, Snoke, Evans, and their col-
laborators King and Beavan at Cam-
bridge University report a technique
for the measurement of seismic phase
velocity at a single site. Conventional-
ly, determination of the phase velocity
requires a widely spaced network that
measures travel times through the
network as a function of frequency. The
large spacing required limits the seis-
mic wavelength that can be used and
hence limits determination of the earth
structure to relatively shallow depths
(<200 km). It is now found that a verti-
cal seismometer and a volume
strainmeter operating at a single site
can, at least in principle, be used to de-
termine phase velocity without any in-
trinsic upper wavelength limit and
without uncertainties introduced by
local heterogeneities and tilts.
Seber, James, Snoke, and Linde have
investigated the application of
maximum entropy spectral analysis to
seismic data. This has the potential
advantage that in contrast to the more
usual Fourier analysis this technique
is well suited to extract the spectral
content from noisy or incompletely re-
corded signals. Calculations using
simulated signals bear out this expec-
tation to a considerable extent, but
some puzzling results were found
which will require further investiga-
tion.
Atomic Physics and Properties of
Materials
Hofmann and Magaritz have consid-
erably extended earlier work on cation
diffusion in natural melts. In addition
to increased understanding of the role
played by diffusion in natural systems,
these new results display effects that
must be relevant to the fundamental
atomic and molecular structure of sili-
cate liquids.
Measurements of diffusion rates in
basaltic melts show that diffusive mix-
ing distances in a partially molten
mantle may be expected to be on the
order of hundreds of meters in 10 9 yr.
These results are used to test a model
for the development of "hot spot"
sources of oceanic islands which has
been proposed by others and for which
mixing by processes other than diffu-
sion is not expected. The limited diffu-
sion lengths appear to preclude the de-
velopment of enriched regions suffi-
ciently large to be of geochemical im-
portance.
Measurements of diffusion in Sr and
Ba obsidian glass also indicate the ex-
tent of diffusive transport as a
geochemical mechanism. These results
show that at metamorphic tempera-
tures of 700°C, transport distances may
be limited to 20 cm in a million years.
This shows that large widely spaced
whole-rock samples may not indicate a
reliable age of metamorphism when
used for Rb-Sr dating, and that small-
scale (~cm) sampling may be required
to ensure the isotopic equilibration re-
quired to establish an age.
These workers give examples of the
wide applicability of these diffusion
data to problems of trace element
geochemistry, igneous petrology, and
to the microscopic structure of liquids.
Hofmann and Brown have developed
a new technique for measurement of
cation diffusion in basaltic melts
wherein the radioactive tracer is ini-
tially located in a sharply defined zone
within the sample rather than being
located on the surface. This is accom-
plished by bombardment of basaltic
glass with 40-Mev deuterons for the
formation of radioactive species such as
22 Na, 4,i Sc, and 5,i Co within the sample.
This approach minimizes convective
mixing during the subsequent melting
118
CARNEGIE INSTITUTION
of the sample and eliminates evapora-
tion of volatile tracers such as sodium
from the surface. This technique is
found to be highly successful and has
been used to confirm and extend the
measurements found using the earlier
technique.
Brown and Pepper continue their in-
vestigations of fundamental problems
in atomic structure. These are ap-
proached by study of the x rays pro-
duced when heavy ions collide at ener-
gies of a few Mev. From the theoretical
point of view it is not clear how the
transient "molecule" formed by the two
ions at the moment of their collision
should be treated, and there are a
number of experimental results that
still require explanation.
These workers report the most clear
example of the formation of a "united
atom" in which both nuclei temporarily
combine as a single center of charge.
The l and m x-ray bands characteristic
of such an atom are clearly observed,
and the limits of these bands change
exactly as predicted when the atomic
number of the target is changed.
In another experiment it is found
that when praseodymium is bom-
barded with potassium atoms, the
praseodymium l shell x-ray lines are
found as expected, as a consequence of
transfer of electronic vacancies from
the k shell of the potassium ion. How-
ever, the remarkable thing is that only
two of the several possible l lines are
seen. The exchange is selective. This is
believed to represent a form of charge
exchange not observed before, and it is
suggested that it is a consequence of
rotational coupling that occurs while
the two colliding nuclei are relatively
distant from one another.
GALAXIES AND STARS
G. E. Assousa, E. T. Ecklund, W. K. Ford, Jr., C.A. Little, C. J. Peterson, V. C. Rubin, N. Thonnard,
K. C. Turner, and M. A. Tuve
Collaborators
B. Balick, B. F. Burke, J. W. Erkes, J. Graham, C. K. Kumar, J. Oort, M. S. Roberts, and R. E. White
Evidence for Motion of the
Galaxy
W. K. Ford, Jr., V. C. Rubin, and N. Thonnard
During the past year we completed
our study, made in collaboration with
Dr. J. A. Graham from Cerro Tololo
Inter-American Observatory and Dr.
M. S. Roberts from the National Radio
Astronomy Observatory*, of a sample
of Sc I galaxies, which was undertaken
to investigate the isotropy of the Hub-
ble expansion (Rubin et at., 1976a,b).
As discussed in detail in last year's re-
port {Year Book 74, p. 113), we have
detected an anisotropy, in the sense
* Operated by Associated Universities, Inc.,
under contract with the National Science Found-
ation.
that galaxies on one half of the sky
have velocities that are small for their
magnitudes; galaxies on the remaining
half of the sky have velocities that are
large for their magnitudes. From the
extensive observational and statistical
study making use of optical mag-
nitudes, optical velocities, 21-cm mag-
nitudes, 21-cm velocities, and diamet-
ers and maximum velocities of the ro-
tation curves, we conclude that a mo-
tion of the Galaxy and of the Local
Group of galaxies relative to the sam-
ple is the only interpretation consistent
with all of the observations. The
Galaxy is moving approximately edge
on, with the leading edge in the anti-
center direction. The velocity of the
center of the Galaxy is V GM ~ 450 ±
125 km s _1 ; as observed at the sun (and
DEPARTMENT OF TERRESTRIAL MAGNETISM
119
therefore including the rotation of the
sun about the center of the Galaxy) the
velocity is V ~ 600 ± 125 km s '.
Alternative explanations of the ob-
served effect which we have considered
but rejected include the following: (1)
Galaxies on one half of the sky are in-
trinsically fainter than those on the
other half. (2) Galaxies observed be-
yond the central region of our Galaxy
are less obscured than those observed
in the anticenter direction. (3) The
Hubble expansion increases by 20%
across the sky for galaxies in the dis-
tance range 3500 < V c < 6500 km s "'.
After correction for the motion of the
observer, the random motions of gal-
axies are small [o-(Au) < 200 km s -1 ],
and the Hubble flow is very uniform
[a(AH/H) < 0.04]. This result thus sup-
ports the similar earlier conclusions of
deVaucouleurs (1958) and Sandage
and Tammann (1975). The conflict
with the observed isotropy of the 3°K
relic radiation, which implies V < 300
km s _1 , remains unresolved.
References
deVaucouleurs, G., Further evidence for a
local supercluster of galaxies: rotation
and expansion, Astron. J., 63, 253, 1958.
Rubin, V. C, W. K. Ford, Jr., N. Thonnard,
M. S. Roberts, and J. A. Graham, The
motion of the Galaxy and the Local
Group determined from the velocity
anisotropy of distant Sc I galaxies. I. The
data, Astron. J., 81, in press, 1976a.
Rubin, V. C., N. Thonnard, W. K. Ford, Jr.,
and M. S. Roberts, The motion of the
Galaxy and the Local Group determined
from the velocity anisotropy of distant Sc
I galaxies. II. The analysis for the mo-
tion, Astron. J., 81, in press, 1976b.
Sandage, A. R., and G. A. Tammann, Steps
toward the Hubble constant. V. The
Hubble constant from nearby galaxies
and the regularity of the local velocity
field, Astrophys. J., 196, 313, 1975.
A Search for Extragalactic
Extinction
V. C. Rubin and N. Thonnard
Knowledge of the type and quantity
of intergalactic material is of interest
because it can help determine the total
amount of matter in the universe and
place limits on the amount of material
outside the galaxies. Attempts to de-
termine this have involved searching
in the direction of galaxy clusters for
excess reddening of galaxies, changes
in the number density of distant
galaxies, and excess luminosity in the
space between galaxies. We have ex-
tended the approach used by Rubin et
al. (1976) in determining the extinction
within our Galaxy to search for excess
extinction between the galaxies in the
direction of the supergalactic plane,
the flattened disk of nearby galaxies
(radial velocities less than 2000 km s _1 )
of which we are an outlying member.
Because we are located in the super-
galactic plane, if extragalactic matter
is distributed in this plane, our line of
sight to more distant galaxies passes
through this layer of obscuring matter.
Taking a uniform sample of distant
galaxies that have a small spread in
intrinsic luminosity (Sc Fs with 3500 <
V < 6500 km s _1 ) and correcting the
observed magnitudes and velocities for
all known systematic effects, one can
form a parameter called the Hubble
Modulus. When we examine the Hub-
ble Modulus as a function of super-
galactic latitude, any systematic
change can be interpreted as a varia-
tion due to the change in path length
through the obscuring material for
galaxies at different latitudes. We find
no variation in extinction as a function
of supergalactic latitude, which implies
that we have detected no intergalactic
material.
An analysis by Takase (1972) of the
corrected colors for 650 galaxies in the
Reference Catalogue of Bright Galaxies
marginally showed that elliptical and
SO galaxies within 10° from the super-
galactic plane are reddened with re-
spect to galaxies at supergalactic
latitudes greater than 30° by approxi-
mately 0.04 magnitude at blue
wavelengths. G. deVaucouleurs et al.
(1972) repeated this analysis and in-
cluded colors for 262 additional
galaxies observed at the 91-cm
120
CARNEGIE INSTITUTION
1 1 1 1 !
i
i i i i_
Sc galaxies:
3500 < V < 6500 -
I.I
-
0<SGL<I80°
•
1.0
•
•
•• •
•
•
•
• -
0.9
•
' * «"
•
i
l ill"
--0.8
-0.4
90 45 40 35 30 25 20 15
I Supergalactic Latitude I (degrees)
E
<d
+0.4
10
Fig. 1. Hubble Modulus for 49 Sc I galaxies at supergalactic longitude less than 180° as a function
of the absolute value of supergalactic latitude. Solid line is least-squares fit to data giving total
change from SG pole to SG equator of -0.07±0.10 magnitudes. Dashed line is from de Vaucouleurs et
al., (1972), using galaxy colors, and gives a total change from SG pole to SG equator of +0.18±0.05
magnitudes.
McDonald reflector. Their results also
marginally indicate excess reddening
in the supergalactic plane for both el-
liptical and spiral galaxies and fur-
thermore indicated a greater effect in
the direction of the Virgo cluster.
If one assumes that the excess red-
dening is due to wavelength-dependent
intergalactic absorption and that the
reddening mechanism for galaxies in
the supergalactic plane is the same as
in the galaxy, then the excess redden-
ing seen by Takase implies an excess
extinction of +0.16 magnitude between
the supergalactic plane and the pole.
However, we see the not significant dif-
ference of -0.12 ±0.08 magnitude. In
Fig. 1 we plot the Hubble Modulus
against the argument of the empirical
reddening law derived by deVau-
couleurs. The solid line is a least-
squares solution to the data, while the
dotted line is the effect found by
de Vaucouleurs et al., converted from
colors to blue magnitudes. Their re-
sults indicated a total extinction of
+ 0.18 + 0.05 magnitude toward the
supergalactic plane, whereas we see
-0.07+0.10 magnitude. Our results
imply an upper limit to the density of
intergalactic material in the Local
Supercluster of 10~ :!I g cm ! , which is
approximately the same as the
minimum average density of the uni-
verse and therefore would at most only
double the density. Therefore, the
"missing matter" needed to close the
universe is not in intergalactic space in
a form that can cause significant ex-
tinction.
References
de Vaucouleurs, G., A. deVaucouleurs, and
H. G. Corwin, Jr., Intergalactic extinc-
tion in the Local Supercluster. I. Selec-
tive extinction, Astron. J., 77, 285, 1972.
Rubin, V. C., W. K. Ford, Jr., N. Thonnard,
M. S. Roberts, and J. A. Graham, The
motion of the Galaxy and the Local
Group determined from the velocity
anisotropy of distant Sc I galaxies. I. The
data, Astron. J., 81, in press, 1976.
Takase, B., Intergalactic reddening of
galaxies, Publ. Astron. Soc. Japan, 24,
295, 1972.
DEPARTMENT OF TERRESTRIAL MAGNETISM
121
The Statistics of Galaxy
Alignments
K. C. Turner
From time to time in the astronomi-
cal literature, processes are inferred on
the basis of the alignment of various
objects. For example, a number of
galaxies are found to lie along a
straight line. Such alignments have
been offered as evidence for the ejection
of small galaxies by large ones. Quan-
titative estimates of the unlikelihood of
such arrangements, however, have
been very rare and not of high quality.
An attempt to understand the probabil-
ity structure of such alignments was
begun some two years ago.
Extensive Monte Carlo calculations
to determine the probability of finding
m objects lying within a given distance
of a straight line when there are N ob-
jects in a square field have been com-
pleted (see Year Book 73, pp. 886-887.)
This approach has been necessary be-
cause the problem is quite intractable
analytically. (A recently published at-
tempt at such a calculation [Barnothy,
1974] is incorrect). Figure 2 illustrates
d/l_\
.005
.01
.02
.03
.04
.05
.06
.07'
.08
.09
.10
N
3,4, 5 | 6 | 7 | 8 , 9 , 10
6
7
the general nature of the results. Given
the number of objects, N, and the ratio
of the half width of the line to the side
of the square field, the areas where one
may find m objects aligned with proba-
bility greater than 10% are labeled
with the value of m. Thus, for example,
in a field of nine objects, the odds are
greater than 1 in 10 that one can find
five objects lying within 5% of the field
size of some straight line. Figure 3 is a
similar representation at the 1% prob-
ability level.
N
a/i_ \
3
4,
5 ,
6 ,
7
8
9.
10
005
.01
.02
3
4
.03
.04
5
.05
.06
6
.07
.08
(
.09
8
.10
6
9
P > 0.10
Fig. 2. Contours of TV for P &10% .
P > .01
Fig. 3. Contours of N for P ^ 1%.
Tables produced by these calcula-
tions have been used to make estimates
of the likelihood that the alignments of
galaxies proposed by Arp (1972 and
references therein) as evidence for the
expulsion of QSOs and galaxies from
galactic nuclei are merely coinci-
dences. In all cases examined, the
probability of finding such alignments
by chance has exceeded 10~ 3 . Figure 4
illustrates one such case, the region of
NGC 520, a peculiar galaxy repre-
sented by the square. All quasi-stellar
objects in the Parkes Catalog in this
region are plotted as open circles if
122
CARNEGIE INSTITUTION
Fig. 4. Quasi-stellar objects near NGC 520.
they lie near Arp's suggested line of
ejection and as filled points if they are
far from the line. The field contains ten
objects, six of which are near the line.
The ratio of line width to the side of the
area searched (calculated from dis-
tance of the farthest open circle to the
line) is 0.03. The probability of finding
such an arrangement by chance is
0.013 (If we eliminate the farthest
point, we can reduce the accuracy ratio
to 0.02, but then only five of the ten
points lie on a line, and the probability
becomes 0.048.)
References
Arp, H. C, Ejection of small compact
galaxies from larger galaxies, IAU
Symp. No. 44, D, Reidel Publ. Co., New
York, p. 380, 1972.
Barnothy, M. F., The NGC 520 chain of
quasars, Astron. J., 79, 8, 1974.
Kinematic Studies of NGC 3351,
3115, and 1275
W. K. Ford, Jr., C. J. Peterson, and V. C. Rubin
A fundamental question of cosmol-
ogy concerns the arrangement of mat-
ter in the universe. Is most of the mass
of the universe contained in galaxies,
or is there a sizable fraction of mass
outside the galaxies? A second question
concerns the significance of the divi-
sion of galaxies into two major types.
Why do some galaxies have flattened
disks, contain gas, dust, and young
stars in significant quantities, and
usually exhibit spiral structure; while
other galaxies are spheroidal, dust and
gas free, generally structureless, and
contain principally older, evolved
stars? Answers to these questions can
be approached by (1) statistical exami-
nation of large galaxy samples, and (2)
studies of masses of fairly nearby typi-
cal galaxies (deduced from a detailed
study of their velocity fields). It is the
latter approach which has occupied
part of our research efforts over the
past few years.
During the last decade, significant
progress has been made in observing
the velocity fields and hence the mas-
ses of spiral galaxies, as measured from
the emission lines in their spectra.
These emission lines arise in the ex-
cited gas in each galaxy. Although the
gas constitutes in general only a small
DEPARTMENT OF TERRESTRIAL MAGNETISM
123
percentage of the mass of the galaxy, it
is assumed that the gas is in dynamical
equilibrium in the gravitational field
of the stellar population, and can thus
be used as a tracer of the underlying
velocity pattern.
For gas-free stellar populations such
as are found in the bars of barred spiral
galaxies or in elliptical or SO galaxies,
velocity data are almost nonexistent
and mass data are meager. During the
past year, we have directed our obser-
vations toward these difficult objects,
and have obtained spectra of suffi-
ciently high quality so that we could
measure the velocities of the stellar
populations from the absorption lines
in the integrated galaxy spectra. We
report here on studies of three such
galaxies: NGC 3351, a barred spiral
galaxy in which for the first time it has
been possible to measure the velocities
of stars in a bar; NGC 3115, an E7/S0
galaxy that had been studied earlier
but with contradictory results; and
NGC 1275, an exceedingly peculiar
galaxy whose nature has long been a
puzzle.
Motions of the Stars and the Excited-
Gas in the Barred Spiral Galaxy NGC
3351
Although numerous studies of mo-
tions in barred spiral galaxies have
been made, such studies have been re-
stricted to the motions of the excited
gas in the nuclei and in the outer spiral
arms. Because of the difficulties in de-
termining accurate velocities from stel-
lar absorption features, no velocities
for the stars in the bars had been ob-
tained. We have now completed a study
of the velocities of both the stars and
the gas in the barred spiral NGC 3351,
and so for the first time can describe
the motions of the stars in the bar of a
galaxy of this type.
NGC 3351 is one of the few bright
barred spirals in the Northern sky
(Fig. 5a). It is a particularly good can-
didate in which to search for noncircu-
lar motions of the stellar population in
the bar because the galaxy is oriented
on the sky with the bar only a few de-
grees from the line of sight. A spectrum
taken at the 4-m telescope with the
spectrograph slit aligned along the bar
is shown in Fig. 5a (right). Prominent
absorption lines arising from the stel-
lar population in the bar are marked. A
careful inspection of the spectrum re-
veals that each galaxy line is red-
shifted — 800 km s _1 (due to the sys-
temic velocity of the galaxy) from a
corresponding but generally weaker
line at V = km s _1 . The latter lines
arise from the scattered solar spectrum
in the night-sky radiation. It is the
high resolution of the spectrum that
permits us to resolve the galaxy fea-
tures from these night-sky features
and to obtain accurate measures for the
galaxy lines. Velocities measured
along the bar indicate that the stars
are moving in circular orbits around
the nucleus, with velocities that in-
crease linearly with distance from the
nucleus. There is no evidence of large-
scale stellar streaming motions in the
bar.
Measures of both stellar and gas fea-
tures in the galaxy lead to the follow-
ing conclusions:
1. The nuclear region contains a
ring of HII regions r ~ 340 pc, which is
rotating with V rot = 126 ± 16 km s" 1
and contracting to the nucleus with V c
= 34 ± 11 kms" 1 .
2. The stars in the bar (r =£ 3 kpc)
are rotating with constant angular ve-
locity 80 ± 20 km s" 1 kpc" 1 . The bar is
thus a quasi-stationary feature in this
galaxy. There is no evidence for large-
scale streaming motions of the stars in
the bar, nor evidence of excited gas in
the bar.
3. Beyond the region of the bar, the
velocities of the HII regions within the
outer ring reach a maximum of V rol ~
220 km/s at R ~ 3-5 kpc and then
begin to decrease.
4. A simple mass model for the
galaxy gives a total mass and angular
momentum consistent with values
found for normal spiral galaxies. Thus
in NGC 3351 there is no indication that
the presence of the bar, so prominent in
optical photographs, distorts the
124
CARNEGIE INSTITUTION
NGC 3351
NGC3115
X
NGC 1275
so
uth
4— — i smmmm t m
•
north
Fig. 5. Absorption line spectra from three different types of galaxies; all spectra taken with the
Kitt Peak 4-m spectrograph, which incorporates a Carnegie image tube, (a) NGC 3351, a barred spiral
galaxy. Spectrum of the bar shows prominent galaxy absorption lines displaced to red from corre-
sponding lines in the night sky. (b) NGC 3115, a spheroidal (E7/S0) galaxy and a spectrum along the
bar. Note the variation in velocity along the disk, indicated by the curvature in the absorption lines.
Galaxy photograph courtesy of S. Strom, KPNO. (c) The peculiar galaxy NGC 1275 and a spectrum
crossing the nucleus. To the south, prominent absorption lines are seen which are obscured to the
north. This is attributed to the presence of an intervening galaxy. Emission lines from both galaxies
(at different velocities) are seen to the north (bottom of spectrum).
large-scale gravitational field in the
galaxy. The surprising result from this
study is that NGC 3351 has a velocity
field that appears normal, much like
that of a typical nonbarred spiral
galaxy.
The Rotation Curve of the E7/S0 Galaxy
NGC 3115
NGC 3115 is a nearby galaxy origi-
nally classified as an elliptical. More
recent photographs have revealed a
flattened disk embedded in a spheroi-
dal system (Fig. 5b). Because of its rel-
atively high surface brightness, its
large angular extent, and the presence
of a disk, it is a prime target for a study
of the dynamics and kinematics of an
elliptical galaxy, albeit a nontypical
one. Earlier spectroscopic studies at-
tempted to determine the rotation of
these galaxies. An early one by Min-
kowski (1960) indicated a minimum in
the circular velocities at large dis-
tances from the nucleus; a more recent
study by Williams (1975) failed to re-
veal the minimum, but his data showed
a large scatter in the measured veloci-
ties.
We have obtained two spectra of very
high resolution along the major axis of
the galaxy (Fig. 5b), using the 4-m
DEPARTMENT OF TERRESTRIAL MAGNETISM
125
Mayall telescope with Carnegie image
tube spectrograph. Measurement of the
positions of 14 absorption features of
Call, Fel, H, Mgl, and CrI as a function
of distance from the nucleus gives rota-
tion curve data with extremely small
statistical errors (Fig. 6). The central
angular velocity is 17.5 ±1.0 km s" 1
arcsec -1 . The velocity curve rises
rapidly, but beyond 20 arcsec it re-
mains level at V ~ 270 km s" 1 . Al-
though we intend to obtain additional
observations at greater distances from
the nucleus before making a final mass
model, we can draw some preliminary
conclusions based on an extremely
simple model. A spheroid of axial ratio
cla = 0.4 and density distribution given
by p = 9 x 10 10 [1 + r/0.2] " 2 ' will re-
produce the flat rotation curve outside
of the nucleus. An additional nuclear
mass of some 10 9 9TC©is implied by the
shape of the rotation curve and the ve-
locity dispersion: The mass out to r = 5
kpc is about 7 x 10 10 3TC©. The projected
surface density of the adopted spheroid
gives a mean surface brightness as a
function of distance from the nucleus
which is too bright near r = 5 kpc com-
pared with the observed surface
brightness (Strom et al., 1976). Hence
the mass-to-light ratio of the inte-
grated starlight must increase by a fac-
tor of 3 or 4 from r = 1 to r — 5 kpc in
the disk. We will attempt to use this
variation, together with the observed
radial gradient in color, to model the
dynamical and chemical history of the
galaxy.
The distance of NGC 3115 is uncer-
tain. It does not appear to be a member
of any clearly definable group of
galaxies, and its systematic velocity of
395 km s" 1 (corrected for the motion of
the sun and Galaxy in the Local Group
of Galaxies) is too small to give an ac-
curate distance estimate by use of the
Hubble law. Minkowski (quoted by
Williams 1975) gives a distance of D —
6.6 Mpc based on the apparent bright-
ness of the globular clusters associated
with the galaxy. We have adopted a
distance of R = 10 Mpc for our calcula-
tions.
in
1000-
£ 800
o
o
_J
LU
> 600
o
or
2 400
o
o
-J
UJ
♦ * * V •*
*♦ .
NE
NGC 3115
SPECTRUM 4M-II26A
PA. 44° (MAJOR AXIS)
SW
-100
-80
-60 -40 -20 20 40 60
RADIAL POSITION (ARC SEC)
80
100
Fig. 6. The observed velocity in the E7/S0 galaxy NGC 3115 as a function of distance from the
nucleus. Each data point represents the mean velocity over a radial range of 5 " from measurements of
stellar absorption lines due to Call, Cal, Fel, Mgl, CrI, and H.
126
CARNEGIE INSTITUTION
The Peculiar Galaxy NGC 1275
NGC 1275 (Perseus A) is one of the
most peculiar objects in our region of
the universe. Twenty years ago, Min-
kowski (Year Book 54, p. 25) discovered
in its spectrum two sets of emission
lines, one set with a redshift of V ~
5200 km s" 1 and a second with V—
8200 km s" 1 . Since that time NGC 1275
has been interpreted as either an
exploding galaxy or a collision between
two galaxies. In order to study the un-
derlying stellar population and the ex-
cited gas in NGC 1275, an extensive
body of spectroscopic and photographic
material has been obtained at the
2.1-m and 4-m telescopes of Kitt Peak
National Observatory. Analysis of this
material in collaboration with Profes-
sor J. H. Oort leads to the following
conclusions:
1. The luminosity profile of the stel-
lar continuum (see Fig. 7) shows that
■
NGC 1275
.
PLATE 4M-I938
■■-.
"■..
"■.
N
KING MODEL CURVE ■
C = 1.5 ''■•■
'■
■ _
1.2
Log r 1
2.0
Fig. 7. The luminosity profile of the con-
tinuum light from NGC 1275, from an east-west
trace. The plate is a 144-min exposure at the
prime focus of the 4-m Mayall telescope, N 2 -
baked 0.98 emulsion + A.6558 filter; FWHM
92 A. In the passband of this plate and filter com-
bination, there is no light from emission in either
velocity system in the galaxy. The ordinate is the
logarithm of the surface brightness, f, of the
galaxy in units of the surface brightness of the
sky. The abscissa is the logarithm of the radius
in seconds of arc. The dashed line is a profile from
the dynamical models of King (1966) with con-
centration parameter C = 1.5. The surface
brightness (solid squares) diverges from the
model curve at small radii (r < 4"( due to an un-
resolved component in the nucleus of the galaxy.
the main galaxy is spheroidal (Fig. 5c).
A strong absorption line spectrum is
seen at V = 5200 km s ', which agrees
with the velocity of the extensive fila-
mentary material surrounding the
galaxy. South of the nucleus is a strong
A-type absorption line spectrum; to the
north the Balmer lines are obscured.
Emission lines of H, O, and N are also
seen (Fig. 5c).
2. Emission patches at V = 8200 km
s -1 over the northern part of NGC 1275
are associated with obscuration that
hides the stellar continuum of the un-
derlying galaxy. The pattern of veloci-
ties in the high- velocity system (Fig. 8)
shows a fairly smooth transition from
velocities near 8100 km s" 1 in the east
to velocities of 8400 km s~' in the west.
This pattern is consistent with that ex-
pected from a late-type galaxy rotating
with a maximum velocity of V r max = 150
km s~Vsin (inclination angle). In addi-
tion, the emission line ratios seen in
this high velocity gas are typical of
large, low-density HII regions observed
at large distances from the nucleus in
late-type galaxies. No stellar absorp-
tion lines have been found associated
with this system. This high-velocity
gas is in front of the low- velocity sys-
tem.
NGC 1275 is at the center of the Per-
seus cluster of galaxies. A count of
galaxies in the central region of the
cluster suggests that the probability of
finding a spiral galaxy within 30" of
the nucleus of NGC 1275 has the rea-
sonably high value of 1/30.
3. NGC 1275 is a strong radio source
as well as a strong infrared and strong
x-ray source. This plus the optical ap-
pearance of the low-velocity fila-
mentary material, reminiscent of the
Crab Nebula supernova remnant in
our own Galaxy, implies that the object
has undergone a violent event in the
past. The A-type stellar spectrum indi-
cates that a burst of stellar formation
was associated with this event. It is
likely that the time scale required for
the dispersal of material into the low-
velocity filament system is longer than
the time scale over which the two
DEPARTMENT OF TERRESTRIAL MAGNETISM
NGC 1275 HIGH VELOCITY SYSTEM
127
N
82808340 840
8108 "» 8122 8167 8 02 8100 8209 8302 8344 8392 8386 837C
8360
' 8236
8144
8167 .
10"
8279.
8399
i
8390
8384
-w
Fig. 8. Observed velocities in the high velocity system in the galaxy NGC 1275. The coordinate
system is centered at the nucleus of the early-type galaxy. High velocity gas is seen only in the
northern section of this peculiar object. The smooth variation in observed velocity from east to west is
consistent with that expected from an almost edge-on galaxy, rotating with a maximum velocity near
200 km s-'.
galaxies may have been in gravita-
tional interaction (the two objects are
still approaching each other); thus, the
presence of the second galaxy cannot
have triggered the explosion in the un-
derlying galaxy. Hence our interpreta-
tion of the NGC 1275 phenomenon re-
quires both an explosive event in the
elliptical galaxy and a high-excitation
late-type intervening galaxy. While it
is unlikely that the two main galaxy
masses are interacting, it is still possi-
ble that an energetic nuclear source in
the underlying elliptical may be the
source of excitation for the high-
excitation spectrum observed in the
late-type galaxy.
Moderately peculiar galaxies are
commonplace in the universe. Severely
peculiar galaxies are less so, but the
attempt to understand them has led
some astronomers to question whether
yet unknown laws of physics govern
their behavior. Our observations of
NGC 1275 show that individual prop-
erties, i.e., the luminosity profile of the
underlying galaxy, the observation of
the northern part, the velocity and
line-intensity pattern of the high veloc-
ity system, all can be understood in
terms of previously known characteris-
tics of galaxies. Such is the progress of
astronomy. Only when detailed studies
reveal characteristics that are outside
the realm of previously observed prop-
erties of galaxies can we properly ap-
peal to an unknown physics.
References
King, I. R., The structure of star clusters
III. Some simple dynamical models, As-
tron. J., 71, 64, 1966.
Minkowski, R., Problems of extragalactic
spectroscopy, Ann. Astrophys. 23, 385,
1960.
Strom, S. E., K. M. Strom, J. W. Goad, F. J.
Vrba, and W. Rice, Color and metallicity
gradients in E and SO galaxies, As-
trophys, J., 204, 684, 1976.
Williams, T. B., The rotation curve of NGC
3115, Astrophys, J., 199, 586, 1975.
An Illustration of Galaxy
Redshifts
W. K. Ford, Jr., C. J. Peterson, V. C. Rubin, and
N. Thonnard
In connection with the observing
program to study the radial velocities
of high-luminosity spiral galaxies, we
have obtained optical and 21-cm radio
spectra of a homogeneous sample of
galaxies. Figure 9 has been constructed
to illustrate the effect of increasing
redshift with distance, using both the
optical Ha and the 21-cm hydrogen
lines. Photographs of the galaxies, all
at the same scale, are copied from the
National Geographic Society-Palomar
Observatory Sky Survey prints.
128
CARNEGIE INSTITUTION
O
II
X
u
Z
o
5
z
u
z
<
I—
</)
-8
-?
U
z >
CN IT)
1= >
UK
O ii
Z >
o o 2 o o o
i o i oo i oo
K rO OO O A. O
oi oo u-> o- So
n n cs ii — ii
O > O > CM >
u o
Z >
■> «-
/
_/
.7
I
. ?"
/ «_/- _.
«= i =? 1 == ~*
u
DEPARTMENT OF TERRESTRIAL MAGNETISM
129
The optical spectra of the Northern
Hemisphere galaxies were obtained
with the 84-inch (2.1-meter) reflector of
Kitt Peak National Observatory, and
spectra of the Southern galaxies with
the 60-inch (1.5-meter) reflector at the
Cerro Tololo Inter-American Observa-
tory near La Serena, Chile. A two-
stage Carnegie image tube was incor-
porated into each spectrograph. The
spectra are reproduced here as nega-
tives; dark features indicate emission.
Wavelength increases toward the
right. In each spectrum, the horizontal
band is the spectrum from the inte-
grated starlight in the nucleus. The
superimposed emission knots arise
from the hot gas in the nucleus and in
regions of ionized gas surrounding
young massive stars in the spiral arms.
The principal emission line is Ha, indi-
cated by the arrow.
The Ha emission arises when an
electron in an excited atom makes a
transition from the n = 3 to the n = 2
orbit. The wavelength of the emitted
light, 6563A, corresponds to 6.563 x
10 5 cm. In contrast, the neutral hydro-
gen emission at 21 cm (a wavelength 3
x 10 5 larger than that of Ha) occurs
when the electron in the hydrogen
atom reverses its spin with respect to
the spin of the nucleus. Because neu-
tral hydrogen is a significant compo-
nent of the interstellar material and
hence of the total mass of an Sc I
galaxy, the 21-cm line can be observed
in these galaxies even though the spin
reversal in an individual atom is a rare
event.
The theoretical resolving power of a
telescope is a measure of the diameter
of the telescope in units of the
wavelength of the observed light; the
size of the 84-inch (2.1-m) is 2.1
m/6.563 x 10" 5 cm = 3 x 10 ,; Ha
wavelengths. In practice, however, the
limiting resolution is set by the turbu-
lence of the atmosphere, and the
resolving power is generally about 1",
or 2 x 10 5 . To observe a galaxy at a
wavelength of 21 cm with the equiva-
lent resolving power would require a
telescope of 21 cm x 2 x 10 5 = 40 km,
or about 25 miles. Lacking this, we
have used the largest single telescope
at the National Radio Astronomy Ob-
servatory, the 300-foot (91-m), which
gives a resolving power of 91 m/21 cm
= 4 x 10 2 , or only l/500th that of the
84-inch telescope. Hence, instead of re-
solving 1" in the galaxy, the radio tele-
scope integrates over about 10' of the
sky.
With the 300-ft transit telescope
equipped with a new low-noise cooled
receiver, we have obtained spectra of
program galaxies with velocities up to
V = 10,800, or AA/A = 0.036. The
galaxy with the largest redshift,
1058 + 11, is the most distant one thus
far for which the 21-cm emission line
has been detected. The galaxy velocity
V is adopted as the midpoint of the in-
tensity profile; the arrow beneath each
profile marks the optical velocity. In
general, the optical and the 21-cm ve-
locities agree well. Where they do not,
the difference might represent a dif-
ference in the velocity of the excited
hydrogen in the nucleus compared with
the mean velocity of the neutral hydro-
gen throughout the galaxy. More
likely, the velocity difference reflects
uncertainties in the optical and 21-cm
velocities. The general agreement of
the two velocities over a wavelength
range of 3 x 10 5 is a stringent test of
the wavelength invariance of the Dop-
pler expression.
The final adopted velocity V for each
galaxy is listed in Fig. 9. With a value
Fig. 9. A new illustration of the Hubble recession of the galaxies, showing both optical and radio
spectra for 13 spiral galaxies. The galaxy photographs are reproduced from the National Geographic
Society-Palomar Observatory Sky Survey. The optical spectra were taken with an image tube spec-
trograph on the 84-in (2.1-m) reflector at Kitt Peak National Observatory or the 60-in (1.5-m) reflec-
tor of the Cerro Tololo Inter-American Observatory near La Serena, Chile. The 21-cm observations
were taken with the 300-ft (91-m) transit telescope of the National Radio Astronomy Observatory.
With increasing distance from the observer, the emission line of Ha in the optical spectra and the
21-cm line in the radio spectra are shifted farther toward longer wavelengths (right).
130
CARNEGIE INSTITUTION
for the expansion parameter ofH = 50
km s" 1 Mpcr 1 , the distance of each
galaxy is found from W50. These dis-
tances are indicated at the top of Fig.
10. Because the value of H is still un-
certain, the indicated distances will
have to be changed if the value fori/ is
revised.
The Irregular Galaxy WLM and
Its Surroundings
N. Thonnard
Irregular galaxies frequently have
neutral hydrogen envelopes that are
considerably larger than their optical
image. Recently, a large band of ex-
tragalactic neutral hydrogen, a few de-
grees wide and extending nearly 180°
across the sky, was discovered by
Mathewson et al. (1974). They named it
the Magellanic Stream, as one end of it
was in the direction of the Magellanic
Clouds. It was soon realized that many
irregular galaxies were close to the
stream, and it was even suggested that
some of the concentrations in the
stream might be "proto-galaxies" that
had not yet formed stars. Observations
made in 1973 with the 100-foot CIW-
IAR* radio telescope in Argentina
(Year Book 73, pp. 918-919) of WLM a
low surface brightness irregular
galaxy, similar in many respects to the
Large Magellanic Cloud but much
smaller in angular size — indicated that
the neutral hydrogen extent is consid-
erably larger than the optical image. In
addition, a large concentration of neu-
tral hydrogen at approximately the
*Carnegie Institution of Washington-Instituto
Argentino de Radioastronomia.
■I5°I8
24
30
36
-z.
o
fee
42
48
_l
o
LU
Q
54
■16 -
12 -
O h l m 24 :
36 23 59 48 59 5812
RIGHT ASCENSION
Fig. 10. Contours of neutral hydrogen integrated flux density in Jy km s ' superimposed on the
optical image of WLM.
DEPARTMENT OF TERRESTRIAL MAGNETISM
131
systemic velocity was seen in the vicin-
ity of WLM, which in retrospect was
realized to be part of the Magellanic
Stream. To study the neutral hydrogen
structure of WLM and to attempt to
understand its relationship with the
Magellanic Stream, it was decided to
make higher resolution and higher
sensitivity observations with the
facilities of the National Radio As-
tronomy Observatory (NRAO).
The WLM galaxy was mapped with
the 300-foot transit telescope at NRAO
on a 6 x 6 arcmin grid. Figure 10 is a
contour map of the integrated neutral
hydrogen distribution of the galaxy
superimposed on a high-contrast copy
of the National Geographic Society-
Palomar Observatory Sky Survey
blueprint. Note the large extent of the
neutral hydrogen contours compared
with the optical image, especially at
the southern end of the galaxy. At the
northern end, there is some suggestion
that the hydrogen contours are dis-
torted westward in the same sense as
the "kink" in the optical image. Pro-
files from cuts through the major and
minor axes are shown in Fig. 11a and
Fig. lib. The rotation of WLM can
-370 -320 -270 -220 -170 -120 -70 -20 30
-370 -320 -270 -220 -170 -120 -70 -20 30 80
VELOCITY (km S" 1 )
Fig. 11. (a)Velocity profiles along the major axis of WLM. The declination is indicated at the right
of each profile, (b) Velocity profiles along the minor axis of WLM. The right ascension is indicated at
the right of each profile. Note that the intensity scale is twice as sensitive as in Fig. 12a.
132
CARNEGIE INSTITUTION
a
S =-14° 30
140' Observations
(22 arc min beam)
-360 -320 -280 -240 -200 -160 -120 -80
S=-I4°30'
300 ft Observations C
(10 arc mm beam)
-360 -320 -280 -240 -200 -160 -120
HELIOCENTRIC VELOCITY ( km s"')
Fig. 12. (a) Position-velocity map at 8 = -14°30' in the vicinity of WLM observed with the 140-ft
telescope, (b) Identical map, but observed with the 300-ft telescope. The intensity scales were nor-
malized on the broad feature at 23 h 45 m .
readily be seen in the velocity shift of
the profiles from approximately —80
km s _1 to -190 km s _1 along the major
axis. The weak signals at - — 200 km
s ' seen at the southern and western
edges are due to the eastern edge of the
Magellanic Stream; the peaks at zero
velocity are from neutral hydrogen in
our own galaxy.
The total neutral hydrogen inte-
grated flux density for WLM is 446 Jy +
km s ] . Using a distance of 2 Mpc to
WLM, (Abies, 1974) the total neutral
hydrogen mass is 4.2 x 10* 9TC Q . From
the velocity width of the profiles and
approximate inclination for the galaxy
derived from the neutral hydrogen con-
tours, the total mass of WLM is 3 x 10 9
9TC . Therefore, WLM is about half as
massive as the Large Magellanic Cloud
but contains more than three fourths
as much neutral hydrogen as the LMC,
indicating it is a rather normal irregu-
lar galaxy.
A 5° x 3° region around WLM was
also mapped with the NRAO 140-foot
*1 jansky (Jy) = lO" 26 W m"- Hz~
telescope on a 15 x 15 arcmin grid to
see the fine structure of the Magellanic
Stream and determine its relationship
with the galaxy. One cut in right as-
cension was also made with the 300-
foot telescope to determine the
minimum size of some of the features.
The analysis of these observations is
just beginning, but the remarkable
similarity between Fig. 12a and Fig.
12b indicates that most features are
completely resolved with a 20-arcmin
beam. Even for the "hot" spot at a =
23 h 53 m , one can estimate from the dif-
ference in peak signal between the two
telescopes that its angular dimensions
are not much smaller than approxi-
mately 30 minutes of arc. Therefore,
the hydrogen in the Magellanic Stream
in the vicinity of WLM is of much lower
concentration and of much larger an-
gular extent than in the galaxy itself.
References
Abies, H. D., private communication, 1974.
Mathewson, D. S., M. N. Cleary, and J. D.
Murray, The Magellanic Stream, As-
trophys. J., 190, 291, 1974.
DEPARTMENT OF TERRESTRIAL MAGNETISM
133
The Nucleus of M31
W. K. Ford, Jr., C. J. Peterson, and V. C. Rubin
There are very few galaxies suffi-
ciently close so that their nuclear re-
gions can be studied in detail. M31, the
Andromeda Nebula, is a major excep-
tion. The limiting resolution generally
available on earth, 1", subtends 3 pc on
M31. (In our Galaxy, at the position of
the sun, a sphere of radius 3 pc con-
tains only about ten known stars; at
the nucleus of M31, a comparable re-
gion contains about 10 (i stars.) High-
resolution photographs of M31 can
supply details of galactic structure
with high spatial resolution. Above the
atmosphere, still higher resolution can
be obtained; a recent Stratoscope bal-
loon flight photographed the nucleus of
M31 with a resolution of 0"2 (Light,
Danielson, and Schwarzschild 1974). A
major result of these observations is an
indication that the position angle of the
major axis of M31 is not constant with
distance from the nucleus but appar-
ently rotates with increasing nuclear
distance. If this apparent change in the
major axis arises from the presence of
absorbing clouds and lanes near the
nucleus, then it has no important rele-
vance to the dynamics of M31. How-
ever, if the observations reflect a true
tilt of the angular momentum vector of
the nucleus with respect to the disk,
this fact may be an important piece in
the puzzle of what relation the nucleus
bears to the structure and dynamics of
the outer disk of a fairly normal
galaxy.
Because the earlier observations
cover only a limited range in r, we have
obtained a sequence of direct narrow-
band red photographs of the nucleus of
M31 with a Carnegie image tube at the
Ritchey-Chretien focus of the 4-m tele-
scope with a large spatial scale (6"7/
mm). We have fitted ellipses to the
isophotes to determine the alignment
of the major axis as a function of
radius. In Fig. 13, we show our data
together with previous measures. In
the region from r — 0"2 to r - 1000",
the position angle of the major axis de-
creases uniformly from PA ~ 70° to PA
~ 40°.
90
<
Cl.
<jO
X
<
O
<
80 -
70-
60-
50-
40
30
M31
THIS STUDY
LIGHT etal. (1974)
JOHNSON (1961)
DE VAUCOULEURS (1958)
LINDBLAD (1956)
HUBBLE (1929)
A
A
AAA
A A
10 12 3
Log f"
Fig. 13. The alignment of the major axis as a function of radius for the Andromeda Nebula.
134
CARNEGIE INSTITUTION
It is not possible to deduce the cause
of this variation from our results. How-
ever, from a similar set of observations
in a different spectral region it would
be possible to separate dynamical ef-
fects from absorption effects because of
the wavelength dependence of absorp-
tion.
References
Johnson, H. M., The nucleus of M31, As-
trophys, J., 133, 309, 1961.
Light, E. S., R. E. Danielson, and M.
Schwartzchild, The nucleus of M31, As-
trophys. J., 194, 257, 1974.
Star Counts in Globular
Clusters
C. J. Peterson
A resurgence of interest in globular
clusters has occurred in the last few
years. A primary factor has been the
totally unexpected discovery (Clark et
at., 1975; Canizares and Neighbors,
1975) that several clusters are the
sources of variable x-ray emission. The
role of the globular clusters in the
study of the Galaxy is important: The
numerous cluster color- magnitude dia-
grams that have recently become
available have permitted a reassess-
ment of the scale of the Galaxy (Harris
1975), and other spectroscopic and
photometric studies are being inter-
preted in terms of the evolution of the
chemical abundances in the Galaxy.
As part of a long-term study of the
globular cluster system {Year Book 74,
pp. 133-135), star counts have been ob-
tained in 27 clusters, using available
photographic material in the plate col-
lections of David Dunlap Observatory,
Hale Observatories, and Ray White at
Steward Observatory. The derived dis-
tribution of stars (log surface density
versus log radius) is shown in Fig. 14
for four representative clusters. The
observed data has been fitted with
theoretical curves from self-consistent
dynamical models (King 1966); these
model curves depend on three parame-
ters. The core radius r c is the radius at
which the observed surface density has
dropped to one half the central value; it
is closely related to the binding energy
of the cluster. The tidal or limiting
radius r, defines the "edge" of the clus-
ter; this limit is imposed upon the clus-
ter by the gravitational tidal field of
the Galaxy during the closest approach
of the cluster to the Galactic Center. A
third parameter indicates the bright-
ness scale of the cluster and may be
either the central surface brightness
m or the total brightness of the cluster
m,. The lack of correlation between r c ,
r,, and either scale factor m or m,
shows that three parameters are in-
deed necessary to define the structure
of any given cluster.
The statistical distributions of these
data quantities, as well as their corre-
lations with other cluster properties
and with the spatial position in the
Galaxy of the clusters, are the result
not only of the formative process of the
Galaxy but also of the subsequent
dynamical evolution of the clusters in-
dividually and within the framework of
the Galaxy as a whole. In principle,
therefore, these data hold clues to dis-
tinguish between the two models that
have been proposed for the origin of the
cluster system. The first, proposed by
Peebles and Dicke (1968), suggests
that the clusters formed very early in
the history of the universe. At a later
epoch galaxy formation occurred and in
that process the clusters became spa-
tially associated with the much more
massive galaxies. The alternative
hypothesis suggests that the clusters
formed coevally with the initial col-
lapse of the material that was to form
the Galaxy. Neither theory, however,
has advanced to the point where either
general or specific predictions of clus-
ter properties have been made.
Any theory of the formation and evo-
lution of the cluster system must be
consistent with the following prop-
erties of the cluster system:
1. The three structural parameters
do not correlate with each other or with
DEPARTMENT OF TERRESTRIAL MAGNETISM
135
CF-1030]
CF-1027 > Star counts
C- 1910 J
J I i I I l_
-0.4 0.0 0.4 0.8 1.2
-0.8 -0.4 0.0 0.4 0.8 1.2
Kron andMayall photoelectric
CTIO Schmidt 64287
C -2631
• Illingworth
' C-76
v C-814 \ Star counts
- □ CTIO Shmidt5l50
Fig. 14. Surface density profiles for a representative set of four globular clusters. The data repre-
sents star counts and concentric aperture photoelectric photometry. The ordinate is the logarithm of
the surface density (stars per unit area in arbitrary units), and the abscissa is the logarithm of the
radius in minutes of arc. The solid line in each figure is a theoretical surface density profile from the
dynamical models of King (1966). Core and tidal radii are indicated as well as the concentration class
(c = log r,lr ( .) for each cluster.
other cluster properties (e.g., color,
metallicity, HR diagram morphology)
which are related to the stellar popula-
tion of the cluster.
2. There is no correlation between
the concentration class (defined as c =
log r,/r ( .) of a cluster and its position in
the Galaxy. There is, however, a gen-
eral tendency for clusters with large
core radii (i.e., weakly bound clusters)
to be absent from the center of the
Galaxy. This is not unexpected, as
these weakly bound clusters would be
preferentially disrupted near the cen-
ter of the Galaxy by the strong galactic
gravitational tidal field. Because the
limiting radius r, is set by the tidal
field of the Galaxy, larger clusters tend
to be found farther from the Galactic
Center.
3. As the radial extent of a cluster is
set by the tidal field of the Galaxy at
that point in the cluster orbit where
the gravitational forces are strongest,
this datum (r t ) can be used to deter-
mine the closest approach of the cluster
to the Galactic Center. This gives a
fifth orbital parameter which, when
combined in a statistical analysis with
the three components for the present
136
CARNEGIE INSTITUTION
position of the cluster and its observed
line-of-sight velocity, yields the distri-
bution of orbital eccentricities for the
cluster system (Fig. 15). This distribu-
tion is consistent with an isotropic ve-
locity distribution (i.e., the dependency
on velocity of the phase space distribu-
tion functionary) goes only as |l>| 2 );
this result has one immediate conse-
quence of great importance. As the
clusters range to great distances from
the Galaxy, they are the best objects to
use for the study of the effect of the
total gravitational field of the Galaxy.
With this conclusion, more stringent
limitations can be applied to the
theoretical analysis of the positions
and velocities of the clusters which will
yield a more accurate estimate for the
total mass of the Galaxy.
0.2 0.4 0.6 0.8
ECCENTRICITY
1.0
Fig. 15. A histogram showing the distribu-
tion of orbital eccentricities computed from the
data for 61 globular clusters. The straight line is
the distribution of eccentricities expected if the
phase space distribution f(r,v) for the clusters is
isotropic in velocity.
References
Canizares, D. R., and J. E. Neighbors, Ob-
servation of a bright variable x-ray
source in a globular cluster, Astrophys.
J. (letters), 199, L97, 1975.
Clark, G. W., T. H. Markert, and F. K. Li,
Observations of variable x-ray sources in
globular clusters, Astrophys, J. (letters),
199, L93, 1975.
Harris, W. E., Distance to the galactic cen-
ter measured from globular clusters,
Bull. Amer. Astron. Soc, 7, 436, 1975.
King, I. R., The structure of star clusters.
III. Some simple dynamical models, As-
tron, J., 71, 64, 1966.
Peebles, P. J. E., and H. R. Dicke, Origin of
the globular star clusters, Astrophys. J.,
154, 891, 1968.
Neutral Hydrogen Supernova
Remnant Study
G. E. Assousa
The survey of old galactic supernova
remnants (SNR) at 21 cm reported ear-
lier (Assousa and Erkes, 1973; As-
sousa, Balick, and Erkes, 1974) was
carried out to study interaction of
SNRs with the interstellar medium,
and in particular to test the hypothesis
that SNRs accrete a shell of interstel-
lar material during the last phase of
their evolution. This is the phase of the
SNR's history during which the rem-
nant expands while conserving
momentum but not kinetic energy. The
onset of this phase is characterized by a
shock velocity of 200 km/s (Woltjer,
1972) and ends when the shock velocity
becomes comparable to the random in-
terstellar gas motions, i.e. —10 km/s.
Although in principle, estimates of
the mass accreted by the SNR can be
made on the basis of the Sedov-Oort So-
lution (Woltjer, 1972), it could not be
calculated without knowing both the
SNR's radius and its expansion veloc-
ity.
Until recently, only two large rem-
nants, Cygnus Loop and HB21, with
filament expansion velocities below
200 km/s., had been measured. To date
we could only verify the existence of
expanding fragmentary neutral hydro-
gen shells in the vicinity of two rem-
nants: HB21 (Assousa and Erkes,
1973) and S147 (Assousa, Balick, and
Erkes, 1974). We have obtained their
DEPARTMENT OF TERRESTRIAL MAGNETISM
137
masses and expansion velocities. We
found marginal evidence for a shell as-
sociated with the Cygnus Loop (As-
sousa, Balick, and Erkes, 1974) and
W41 (Assousa, 1975). A detailed study
of 3C396.1, W50, CTB 63, W63, CTB 1,
HB 3 and HB 9, and PKS 0607 + 17
shows no evidence for any shell struc-
ture. The remaining objects CTB 72,
CTA 1, CTB 13, VRO 4205.01, and OA
184 have been partially analyzed, but
even for these objects, with the excep-
tion of VRO 4205.01, there appears lit-
tle evidence for HI which we are able to
detect with present instrumental sen-
sitivity (peak-to-peak noise 0.3° K).
The absence of these shells, initially
puzzling, becomes understandable, at
least for some remnants, with the
availability of filamentary velocity
data (Lozinskaya, 1975). Knowledge of
these velocities and the radii of the
SNR allows us to make the mass esti-
mates. Table 1 lists in columns 1
through 9 the object name, longitude
and latitude of object, distance to object
(R) in kpc, filament or shell velocity (V)
in km/s; radius (r) in pc; interstellar
density (p ) in H atoms/cm 3 , accreted
mass (3TC) in 911 , age (t) in 10 3 yr, and
initial explosion energy (EJ in 10 50
ergs.
Given R and V, the interstellar
medium density p is uniquely defined
by the Sedov (adiabatic-energy con-
serving phase)-Oort (isothermal-
momentum conserving phase) solution
(Woltjer, 1972) for an initial explosion
energy of 1 x 10 50 ergs. The family of
solutions forp = 0.01, 0.1, 1.0, and 10.0
H atoms/cm 3 is given in Fig. 16.
Examination of the mass column shows
io 3
10°
lO* 10-
TIME (YEARS)
Fig. 16. The evolution of a 1 x 10 5 " erg SNR. The family of curves represents the expansion
velocity vs. time (double-slope lines, decreasing), radius vs. time (double-slope lines, increasing), and
accreted mass vs. time for a range of ISM densities.
138
CARNEGIE INSTITUTION
PS
CO
O
w
CO CM
o
O O
i— 1 r- 1 CO
CO ^
o
o o o
o m
■* cm in
CD ■<*
t-H Tf T?
T t
o o o o o
f"tONON
(N rH ■<* iH
O l>
O CN
^ Tf S?
co co
tH O
in p oq co ©
O r-i O O r-i
• m • -co
. OQ
O CO i— I I-H T— I
OKffi >C0
£>
in »
CN ^
o I C#
INOOlOO
inskaya,
vey.
dov-Oort
ents that
H(M D- C005
t> O t-h o o
OHH^H
8 5% 6
CD t> CO CO t~
CO Tt OJ Tt T-H
OHU3NO
ned from L
ned from s
ined from
11 detected,
no HI fila
rfOJOtOO
t> CO CD <X> 00
+5 j$ wQ
fl-S
^ cosiz;
DEPARTMENT OF TERRESTRIAL MAGNETISM
139
good agreement with measured and es-
timated mass where available and
shows a mass too low to be detected for
the remaining objects, consistent with
our results. The list represents objects
observed in the survey for which veloc-
ities are available.
Lozinskaya (1975) has suggested
that a homogeneous class of objects ex-
panding into a medium of variable
density may account for the observed
properties of a sample of remnants she
has studied. Our observations are con-
sistent with this suggestion. More pre-
cisely, we see that 10 50 ergs is a good
estimate of the initial energy.
We thus summarize that present ob-
servational data for old galactic SNRs
may have arisen from nearly identical
initial conditions (presupernova mass
and SN kinetic energy 10 50 ergs) but
expanded into regions of varying in-
terstellar medium densities.
References
Assousa, G. E., Search for shell structure in
association with old galactic supernova
remnants in neutral hydrogen and for-
maldehyde, Carnegie Inst. Wash. Year
Book 74, 126, 1975.
Assousa, G. E., B. Balick, and J. W. Erkes,
Supernova remnants, Carnegie Inst.
Wash. Year Book 73, 899, 1974.
Assousa, G. E., and J. W. Erkes, An ex-
panding shell of hydrogen surrounding
the supernova remnant HB 21, Astron.
J., 78, 885, 1973.
Lozinskaya, T. A., Supernova remnants as
optical nebulae, Sou. Astron., 19, no. 1,
21, 1975.
Woltjer, L., Supernova remnants, Ann. Rev.
Astron. Astrophys., 10, 129-158, 1972.
Type I Supernovae:
Interpretation of Optical
Spectra
G. E. Assousa, W. Kent Ford, Jr., C. J. Peterson,
and V. C. Rubin
Supernovae (SN) represent an end-
stage of stellar evolution. An under-
standing of this phenomenon is critical
for the determination of the rates at
which SNs contribute to the synthesis
of elements, to cosmic rays, and to the
energy released into the interstellar
medium.
Observationally one is hampered by
the relatively low rate of SN produc-
tion. Ilovaisky and Lequex (1972) es-
timate on the basis of a galactic survey
of supernova remnants a rate of
(50 ±25 years/SN).
For the optical astronomer the
spectra of SNs and their evolution with
time, following initial detection, pro-
vide one observational handle. An un-
derstanding of the nature and the
time-behavior of these spectra may not
provide direct evidence of heavy-
element (more massive than oxygen)
synthesis. It may do so indirectly by
identifying those elements which give
rise to the spectrum. Simultaneously,
it provides information regarding the
physical conditions in the SN which
give rise to these lines.
To date, the problem of line identifi-
cation remains only partially solved.
This is particularly true for supernovae
of type I (SNI). This problem is aggra-
vated by the lack of data in the litera-
ture. While many spectra of super-
novae exist and have been used for
classification of their spectral type,
most spectra have never been pub-
lished. Published spectra exist for only
about 10% of recorded supernovae (Oke
and Searle, 1974).
There are, of course, difficulties in
obtaining supernovae spectra. Few are
bright at the time of discovery. Their
rapid decline in brightness severely
limits the time over which spectra may
be easily acquired; in addition, nearly
one half of the possible observation
time is lost due to the brightness of the
moon-lit sky. The development of fast
image tube systems in the past decade
has made practical the recording of
spectra of faint supernovae. As part of
a continuing program on the study of
extragalactic systems, two of us (VCR
and WKF) have made a practice of se-
curing spectra of supernovae whenever
140
CARNEGIE INSTITUTION
possible (Rubin and Ford, 1967; Ford
and Rubin, 1968).
Here we present the spectra of three
supernovae, SN 1970j in NGC 7619,
SN 1971 1 in NGC 6384, and SN 1971 p
in MCG 5-26-14.
Based on the present and other
available data on the spectra of SNIs,
we have generated synthetic spectra
that closely resemble the observed
spectra.
The original spectra and microden-
sitometer tracings are shown in Figs.
17 through 19. In Fig. 20, the spectrum
of SN 1971 I (C-47) at 18 days after
maximum has been replotted. The bot-
tom plot is a synthetic spectrum com-
posed of emission time of Fell, Call,
Nal, H u , H/i, H y , H«, and the Hell R,
line at 4686 A. There are 48 lines due
to Fell, 2 lines each for Call and Nal, 4
lines for HI, and a single line for Hell.
The wavelength is the rest frame of the
galaxy. The lines have been Gaus-
sian-broadened to reflect a velocity
dispersion of 3000 km/s. For all except
Hell, the relative strengths of the lines
were obtained from the relative line
widths in the Nova DQ Hercules (Mus-
tel, 1974). The Hell contribution was
arbitrarily chosen at twice Ha. Al-
though the C-47 spectrum is in density,
the marked similarity of the actual and
synthetic spectra is very striking. In
the synthetic spectra the Fell is the
dominant contributor. Kirshner et al.
(1973) have shown that the peak at
4600 A gradually shifts with time. A
reasonable interpretation is that the
Hell 4686 line becomes more promi-
nent with time, becoming considerably
stronger than the Fell.
Improved spectral fits may be ob-
tained by including other lines ex-
pected to be present, such as Sill, but
the improvement remains small.
We are thus led to conclude that
spectra of SNIs are primarily domi-
nated by Fell transitions. We have not
attempted to devise a complete model
to determine the physical conditions in
the SN which give rise to this result. A
radioactive excitation mechanism by
resonance transitions at 2500 A, fol-
lowed by permitted transitions to lower
metastable states as suggested by
Wampler and Oke (1967) for identify-
ing the Fell lines in the QSO 3C273 is
physically plausible for SNIs.
But, regardless of the mechanism,
the presence of Fe in SNI appears ines-
capable.
REST W AVELE NGTH
(A)
3800 4200 4600 5000
5400 5800
6200
6600
6
SN I970j
ton
■
5
1943
Hg
1
:
4
>-
H 9 iii i>VVr\
■
h-
1 /VU, ff \k A
Hg
I
NoD
■
2 a
a J i il*» ' wl r^/vAi a
1
1
■
UJ
1-
2 1 v V 1 III J
\ vr
con
? 2
1
^Vv-vngvvi
\ Hvni
A
1
iVtJw 1 i i i 1 1 1— 1 1
i i i i
\r^
V
4600
OBSERVED
5000 5400
WAVELENGTH
5800
(A)
Fig. 17. Spectrum of supernova 1970j in the galaxy NGC 7619 at 1 day before maximum light. The
bottom scale gives the observed wavelength; the top scale gives the wavelength in the rest frame of
the galaxy. No corrections have been made for change of instrumental sensitivity with wavelength.
Original dispersion is 57 A/mm.
DEPARTMENT OF TERRESTRIAL MAGNETISM
141
REST WAVELENGTH
4400 5000 5600
6800 7400
SN 19711
C-28
NIGHT SKY LINES
[OT] No D [Oil
4400
OBSERVED
5000 5600
WAVELENGTH
Fig. 18. Spectra of supernova 1971 in the galaxy NGC 6384. Spectrum C-28 was obtained on 29
June 1971 at maximum light (Barbone^ ai, 1973). Spectrum C-47 was obtained 18 days later on 17
July 1971. Because no calibrations exist for these photographic spectra, the data are presented in
terms of density. The top and bottom scales are galaxy rest-frame and observed wavelengths, respec-
tively. No corrections have been made for change of instrumental sensitivity with wavelength. Origi-
nal dispersion is 300 A/mm.
References
Barbon, K., F. Ciatti, and L. Rosino, Obser-
vations of five supernovae 1970-71,
Memoria della Societa Astronomica
Italiana, 44, 65, 1973.
Branch, D., and B. Patchett, Type I super-
novae, Mon. Notic. Roy. Astron. Soc,
161, 71, 1973.
deVaucouleurs, G., and A. deVaucouleurs,
Reference Catalogue of Bright Galaxies,
(Austin: University of Texas Press),
1964.
Ford, W. K., Jr., and V. C. Rubin, The spec-
trum of the 1968 supernova in NGC
2713, Proc. Astron. Soc. Pac, 80, 466,
1968.
Ilovaisky, S. A., and J. Lequex, A study of
galactic supernova remnants, Astron.
Astrophys., 20, 347, 1972.
Kirshner, Robert P., J. B. Oke, M. V.
Penston, and L. Searle, The spectra of
supernovae, Astrophys. J., 185, 303,
1973.
Mustel, E. R., Analysis of the physical con-
ditions inside the envelope of DQ Her-
cules 1934, Vistas in Astronomy, vol. 16,
(A. Beer, ed., Pergamon Press, New
York), pp. 260-269, 1974.
Oke, J. B., and L. Searle, The spectra of
supernovae, Ann. Rev. Astron. As-
trophys., vol. 12, 315-329, 1974.
Rubin, V. C, and W. K. Ford, Jr., The spec-
trum of the 1967 supernova in NGC
142
CARNEGIE INSTITUTION
3400
REST WAVELENGTH (A)
3800 4200 4600 5000 5400
5800
6200
SN I97lu.
2II6B
SN I97lu.
2I20A
3400
4200 4600 5000
OBSERVED WAVELENGTH
5400
(A)
Fig. 19. The spectra of supernova 1971u in the anonymous galaxy MCG 5-26-14. Spectra 2116 B
and 2120 A were obtained on successive nights (16 and 17 December 1971, respectively) and are
consistent with type I SN spectrum at 15±2 days after maximum light. The contributions from the
night sky have been removed from both spectra. As in the previous two figures, no corrections have
been applied for instrumental sensitivity with wavelength. The bottom scale gives the observed
wavelength; the top scale gives the observed wavelength; the top scale gives the wavelength in the
rest frame of the Galaxy. Original dispersion is 109A/mm.
3389 and H-alpha velocities in the
Galaxy, Proc. Astron. Soc. Pac, 79, 322,
1967.
Rubin, V. C, W. K. Ford, Jr., N. Thonnard,
M. S. Roberts, and J. Graham, The mo-
tion of the Galaxy and the Local Group
determined from the velocity anisotropy
of distant Sc I galaxies, I. The Data, As-
tron. J., in press, 1976.
Wampler, E. J., and J. B. Oke, The emission
line spectrum of 3C273, Astrophys. J.,
148, 695, 1967.
DEPARTMENT OF TERRESTRIAL MAGNETISM
143
i — i — i — i — i — i — I — i — i — i — i — i — r
~i i i r
3600 4000 4400 4800 5200 5600 6000 6400 6800
REST FRAME WAVELENGTH IN A
Fig. 20. The SN 1971 / spectrum at 18 days after maximum (top) and the synthetic spectrum
(bottom). The lines used in obtaining the spectrum were broadened with a Gaussian assuming a
velocity spread of 3000 km/s. Positions of the NAI, Call, and Ha,/3,y,8, lines are indicated. The
presence of the Balmer lines produces minor changes in the shape of the spectrum. The Hell line
at 4686 clearly has the effect of redshifting the 4600 A peak. The dashed spectrum is due to Fell
alone. All wavelengths are in the rest frame of the Galaxy.
The 21-cm Receiver for the
Instituto Argentino de
RADIOASTRONOMiA
E. T. Ecklund, C. A. Little, N. Thonnard, and
K. C. Turner
The joint proposal to modernize the
21-cm receiver at the IAR and to
undertake an investigation of high-
velocity clouds and southern spiral
galaxies at the neutral hydrogen line
was funded as of November 1, 1975, by
the National Science Foundation and
the Agencia Ejecutiva Argentina. We
have already had a four-week visit in
January by Ing. E. Filloy, the chief en-
gineer of the IAR, during which we
finalized all of the details in the re-
ceiver design.
Most of the components for the
front-end of the receiver have been re-
ceived or are on order. A block diagram
of the receiver front-end package is
shown in Fig. 21. The parametric
amplifier has been received, and using
the noise figure and/or losses in the
other components, we expect a total
receiver-noise temperature of 56°K.
This leads us to expect a system tem-
perature of 70°K on the sky. The use of
dual frequency conversion allows a
large separation between signal and
image frequencies and simplifies the
design of band rejection filters.
Figure 22 is a block diagram of the
"back end" of the receiver. By utilizing
an on-line computer for phase detection
and integration, we hope to avoid many
nonlinearities and temperature-
dependent effects which proved
troublesome in the old system. In addi-
tion, we will be able to automatically
calibrate the system gain and baseline
every few seconds, thus eliminating
several other possible instability prob-
lems. All components for the digital
part of this system are now on hand,
and their assembly into a working sys-
tem is under way.
144
CARNEGIE INSTITUTION
lAtten.
~l
Vorioble
Noise I* Variable current source
Source
Precision constant
current generator
First L.O.
Mixer
Parametric
Low loss Amplifier
line for
future switch
Transistor
Amplifier
Temperature controller
25°C ± 0. I
L_
Second
H6 L.O.
J
v To receiver back end
Fig. 21. Front end for the new 21-cm receiver at the IAR.
820-MHz Survey of the Southern
Sky
N. Thonnard and K. C. Turner
A recent development in radio as-
tronomy, pleasing both aesthetically
and scientifically, is the possibility of
constructing maps of the whole sky
with respectable angular resolution
(i.e., —1°). Such pictures are revealing
structures so large in angular size as to
be heretofore unobservable.
One such survey of local neutral hy-
drogen by Heiles and his collaborators
at the University of California, Berke-
ley, is now being completed at the IAR
in cooperation with our Argentine col-
leagues.
We are presently designing a system
to produce such an all-sky picture of
the continuum radiation at 820 MHz in
a cooperative program with members
of the IAR and Professor H. Weaver's
group at Berkeley. The northern por-
tion of the survey already exists down
to declination —7° by Berkhuijsen
(1972), done at Leiden. The southern
portion will be done at the IAR, and a
central region overlapping both halves
to ensure proper calibration of scales
will be done at the Berkeley Hat Creek
Observatory. Both polarizations of the
incoming radiation will be measured.
Design and construction of the
back-end and data-acquisition systems
will be done at DTM, while the Berke-
ley group will be responsible for the
front-end system. All of the digital
equipment has already been acquired
and is now being assembled. A KIM-1
microprocessor based system made by
DEPARTMENT OF TERRESTRIAL MAGNETISM
145
->
64
Channel
Analog
to
Digital
Converter
Data
Acguisition
Computer
32 K Bytes
~fc
Present
Filter Spectrometer
30 channels of 100 kHz BW
56 channels of 10 kHz BW
Teletype
Terminal
Tape Rec
Interface
D to A
for
Display
Auxi liary
Interface
D toA
for
Chart
9 -track
Digital
Tape Rec
^Oscilloscope
-~>
Strip Chart
Recorder
-> to Dicke Switch
-> to Noise Source
Sidereal time, hour angle, declination
and local oscillator frequency inputs
Fig. 22. Phase detection, integration, and control system for line spectrometer at the IAR.
MOS-Technology will serve as the cen-
tral control for the system. The
availability of such sophisticated data
handling systems at very low prices
makes possible much more adequate
control of the gain and baseline offset
of the receiver, as these parameters can
be measured automatically every few
seconds.
Reference
Berkhuijsen, E. M., A survey of the con-
tinuum radiation at 820 MHz between
declinations -7° and +85°. I. Observa-
tions and reductions, Astron. Astrophys,
Suppl, 5, 263, 1972.
Image Tube Systems
W. K. Ford, Jr.
The combination of the Carnegie
Image Tube and the Ill-aJ recording
emulsion continues to be used to pro-
duce astronomical data of high quality.
The tubes, which are manufactured by
RCA (Type C33063EP3), were origi-
nally developed with the support of
grants from the National Science
Foundation. These improved versions
of the original Carnegie Tube are now
in operation at a number of obser-
vatories including Kitt Peak, Cerro To-
lolo, and Las Campanas.
During this report year two image
tube systems for spectroscopic work
were built in the Department of Ter-
restrial Magnetism's shop. One of these
systems was designed in collaboration
with Drs. Searle and Brucato of the
Hale Observatories for use with a
Bowen-type f/1.4 Schmidt camera on
the Boiler and Chivens spectrograph at
the new Irenee du Pont Telescope at
Las Campanas. The second system was
built for use on the DTM spectrograph
at Lowell Observatory. It replaces the
early Carnegie Image Tube System
(incorporating RCA tube type C33011)
that has been in regular operation at
Lowell Observatory, especially by Drs.
Rubin and Ford, since 1966.
146
CARNEGIE INSTITUTION
We anticipate that much of our as-
tronomical work in the next few years
will continue to be done with photo-
graphic recording of the image inten-
sifier output. In spite of the problems of
accurate photometric calibration of the
plates on one hand and the great prog-
ress that is being made with direct re-
cording of data with digital sensors on
the other, there are some types of ob-
servations for which the size (or
number of picture elements), stability,
and efficiency of the present intensifier
systems still give an overwhelming ad-
vantage. Therefore, we are undertak-
ing a modest renovation and modern-
ization of the DTM scanning mic-
rophotometer. The principal mechan-
ical improvement is the replacement of
the old traveling stage with a precision
transverse stage of one micron resolu-
tion and driven by a stepping motor.
Also, the card-punch recording of the
data is being replaced by a digital
magnetic tape recorder. These changes
will ease the task of extracting relative
intensity information from the plates.
The intensifier-image dissector
scanners at Lick Observatory and at
the Kitt Peak National Observatory
and the Silicon Intensified Target (SIT)
systems at Hale, Kitt Peak, and Cerro
Tololo Observatories are operating
successfully and are yielding photo-
metric astronomical data with sky
background subtraction. However,
these types of systems appear to fall
short in performance over what might
be achieved with some silicon detectors
now in the development stage. In these
detectors an image is read out serially
point by point by moving photogener-
ated charge laterally through the bulk
silicon of the device to an output gate.
The charge can be accumulated during
an exposure or integration period and
then moved by manipulating in appro-
priate phase the potential applied to
metallic electrodes insulated from the
bulk silicon. These structures are gen-
erally described by the generic term
charge coupled devices, or more fre-
quently simply CCDs.
The basic advantage of the CCD over
more conventional camera tubes that
astronomers seek to exploit is the in-
herent geometrical stability, compact-
ness, and long-term stability of sen-
sitivity. There are two broad ap-
proaches: the photon-in mode, analo-
gous to the silicon target vidicon; and
the electron-bombardment mode,
analogous to the SIT vidicon. The
photon-in mode has the advantage of
high quantum efficiency, a high
signal-to-noise ratio for high signal
levels, and probably a higher dynamic
range than conventional camera tubes.
The relative lack of problems as-
sociated with electron beam-landing on
the target should simplify obtaining
quantitative information. The elec-
tron-in, or event-counting, mode may
eventually offer advantages at very
low light levels with the quantum ef-
ficiencies associated with photoemis-
sive surfaces. The commercial devel-
opment of prototypes of both these de-
vices is proceeding along several com-
petitive lines.
As a result of the long program of
development and evaluation of image
tube systems at DTM we are in a good
position to analyze and evaluate the
characteristics of some of the more
promising of these devices. In particu-
lar, the data handling system we have
used with our image dissector scanner
can be adapted to measuring the per-
formance of the CCDs as they become
available. This preliminary evaluation
can be done without the expensive de-
velopment of a full data handling sys-
tem. As a step toward making some in-
itial tests at DTM and in order to pro-
vide some on-line data compression for
spectroscopic applications, Ford and
Thonnard, assisted by a very capable
summer student, Mr. Charles Bennett,
have designed and built a fast arithme-
tic logic unit to operate with our signal
averaging device. This unit allows us
to sum repetitive scans of an image,
thereby improving the signal-to-noise
ratio and at the same time reducing the
total number of picture elements that
DEPARTMENT OF TERRESTRIAL MAGNETISM
147
need to be stored at the expense of re-
duced resolution in one dimension.
Acknowledgments
We thank the Directors of the Cerro
Tololo Inter-American Observatory,
Kitt Peak National Observatory, and
National Radio Astronomy Observat-
ory for making telescope time avail-
able, and the Director of the Hale Ob-
servatories for granting permission to
reproduce the National Geographic
Society-Palomar Observatory Sky
Survey prints. One of us (CJP) wishes
to thank Drs. R. Racine and S. van den
Bergh of David Dunlap Observatory
and H. Swope of Hale Observatories for
their hospitality during visits to these
institutions. We also wish to thank the
Office of Naval Research for the con-
tinuing loan of the lead ballast used in
the counterweights of our Derwood
telescope (Contract No. N0014-71-C-
0211).
each of one month duration. These
trips are intended for personnel of the
IAR to cooperate in the design, assem-
bly, testing, and data processing for the
joint project.
Five parametric amplifiers at 21 cm
have been received from the Nether-
lands Foundation for Radioastronomy.
These amplifiers were used in the Wes-
terbork Synthesis Radiotelescope dur-
ing its earlier observations. They are
going to be used in the near future in
the radio interferometer of the IAR,
which is planned as an aperture syn-
thesis instrument.
The scientific staff has increased this
year with two young members: Drs. T.
E. Gergely and M. Zales de Caponi,
both of whom received the Ph.D. at the
University of Maryland.
Since July, E. Bajaja has been the
Director of the IAR.
Technical Activities
Activities of the Instituto
Argentino de Radioastronomia
For the IAR as well as for most of the
Argentine observatories, 1975 was a
difficult year because of the economic
situation. Nevertheless, some major
projects have been carried out. One of
them was the relocation of the 21-cm
line receiver to a new control building.
Except for the necessary interruption
during the movement of the receiver,
observations have continued an aver-
age of 16 hours a day.
The joint project between CIW and
IAR for the new 21-cm receiver was
started after the agreement between
the Ajencia Ejecutiva Argentina and
the National Science Foundation was
finalized. Under this agreement the
Consejo National de Investigaciones
Cientificas y Tecnicas de la Republica
Argentina will support the expenses
for three round trips to Washington,
The main task of the Technical De-
partment this year was the installation
of the 21-cm receiver in the new build-
ing. This building was built between
the two 30-m dishes and equipped with
an effective air conditioning system.
There is enough room for the receivers
and for the operation facilities.
The electrical wiring, cable ducts, air
conditioning, lighting, and appliances
were ready around March. The move-
ment of the receiver started in April.
At the same time, some parts were re-
placed with new ones designed and
built at the IAR; for example, the low
frequency section of the LO and the old
SOLA AC regulators were replaced by
small units with harmonic rejection
and better stability. A complete set of
interconnected boxes allows easy
maintenance of the whole system; and
individual power distribution was
added to each rack. Some of the old IF
amplifiers were replaced by new ones
with better stability. A more reliable
temperature controller was installed in
148
the filter boxes, and some of the chassis
were rewired and readjusted. First
tests of the system started in June, and
in July the receiver was put into nor-
mal operation.
The new set-up is much more reli-
able, and stability has been improved
by a factor of 5, allowing a better
front-end adjustment. The system-
noise temperature is about 180°K.
At the end of the year some work was
also done on the second 30-m dish, such
as painting, and construction of cable
ducts, interconnection house, and part
of the electrical installation.
Summary of Current Research
E. Bajaja, F. R. Colomb, and R. Mor-
ras have extended their observations at
intermediate galactic latitudes and are
now analyzing the results of the region
200 ssZ ^315; -10 ^b ^ -32.
Bajaja, Colomb, G. Dubner, and T.
Gergely are completing the study of
four supernova remnants: G 296.3 + 10,
G 261.9 + 5.5, G 327.6 + 14.5 and G
330.7 + 15. Other supernova remnants
of small angular diameter have also
been observed.
Colomb, W. Poppel, and C. Heiles of
the University of California continued
the observations for the hydrogen sur-
vey at low velocities (-50 =5 V *£ 50
CARNEGIE INSTITUTION
km/s) of the Southern Hemisphere.
They completed about 60% of the ob-
servations and produced a radio map of
the Southern sky every 2° in 8.
Colomb and I. Mirabel observed
emission profiles in the direction of 23
southern pulsars.
Bajaja, Colomb, and Poppel began
observations of spiral and irregular
galaxies with m< 9. Thirty wide filters
(AV = 25 km/s) and the multichannel
integrator have been used in this pro-
gram.
Poppel and M. Franco observed the
region 0^ / ^ 20; -3° 5* b ss -17° every
1° in I and b, and continued with the
analysis of the region 348" «s I «£ 372°;
3° =£ b *£ 17° (Atlas of Galactic Hydro-
gen, Poppel and Vieira, CIW, 1974).
Poppel and D. Vota completed the
observation of the galactic clusters
NGC 5460, Rn 106, and G 394.
E. Arnal completed the observations
of the galactic clusters NGC 2243,
3688, 2345, 2204, 4755, 3680, 6611,
and 6994, IC 2391, Mel 227, and A
(Blanco) as a general survey of HI in
the direction of galactic clusters.
Gergely and Arnal completed the ob-
servations of the globular clusters B46,
B67a, B87, and B63.
Arnal completed the observations of
the association CMaOBl.
Gergely and Mirabel completed the
observations of globular clusters with
known reddening.
DEPARTMENT OF TERRESTRIAL MAGNETISM
THE EARTH AND THE SOLAR SYSTEM
149
L. T. Aldrich, C. Brooks, J. Cleary, J. R. Evans, M. Feigenson, S. E. Forbush, S. R. Hart, A. W.
Hofmann, D. E. James, A. T. Linde, M. Magaritz, S. Rajan, I. S. Sacks, M. Schneider, C. M. Seber,
J. A. Snoke, K. Suyehiro, D. J. Whitford, and S. Weidenschilling
Collaborators
C. Aguirre, J. Bannister, R.J. Beauan, K. Bloomfield, M. Bracamonte,R. Cabre,A. Cuyubamba, G. L.
Davis, S. del Pozo, D. W. Evertson, A. Flores, L. Gelinas, A. A. Giesecke, N. Girardin, G. C. P. King,
Y. Motoya, L. Ocola, H. Okada, G. Olafsson,R. W. Page, G. Poupeau, A. Rodriguez B., D. Simoni,
R. Stefansson, S. Suyehiro, L. Tamayo, J. G. Williams, Y. Yamagishi, and E. Zinner
Gas-Solid Interactions in the
Solar Nebula
S. J. Weidenschilling
It is widely believed that the planet-
ary system was formed from matter in
a cloud of gas and dust which sur-
rounded the sun. The origin and struc-
ture of this solar nebula are disputed,
but it is generally depicted as a rotat-
ing, flattened disklike structure. It is
reasonable to assume that it was
somewhat centrally condensed, with
the gas pressure and density in the
central plane decreasing with distance
from the axis. Such a pressure gradient
would partially support the gas against
the sun's gravity, requiring its rotation
rate at any radius to be less than the
local Kepler orbit velocity.
Solid bodies are not supported by the
pressure gradient; in the absence of
gas, they would pursue Kepler orbits.
A balance of gravitational, centrifugal,
and drag forces requires that a particle
always have some velocity relative to
the gas. The radial component of its
motion is always in the direction of in-
creasing pressure, i.e., inward. We de-
fine the "stopping time," t e , as the time
for a particle's velocity to be reduced by
a factor e by drag in the absence of
other forces; t e depends on the prop-
erties of the particle and the gas, and
on the velocity. The particle motion is
characterized by the ratio of t e to the
orbital period, t p . Two limiting cases
were examined quantitatively by
Whipple (1973);
1. Small body (t e /t p « 1). The parti-
cle is constrained to move with the an-
gular velocity of the gas. It feels a re-
sidual central attraction. In a frame
rotating with the gas, it falls directly
inward. The terminal velocity in-
creases with particle size.
2. Large body (t e /t p » 1). The parti-
cle pursues a Kepler orbit, which is
perturbed by drag. It loses angular
momentum and spirals inward. The
larger the body, the smaller is the ef-
fect of drag.
In the limits of very large and very
small bodies, the radial velocity ap-
proaches zero. Somewhere between
these cases (when t e lt„ ~ 1), it must
reach a maximum, as shown schemati-
cally in Fig. 23. This case, potentially
the most interesting, had not been
examined in detail before the present
study. In general, the equations of mo-
tion for the strongly perturbed case
must be solved numerically. However,
some interesting properties of the solu-
tion can be shown analytically:
1. There is an upper limit to the ra-
dial velocity, which depends only on
the nebular structure. It is indepen-
dent of the drag law and the particle
properties.
2. The radial and transverse compo-
nents of the "wind" velocity felt by the
particle are uniquely related. They also
depend only on the nebular structure.
3. If the gas is assumed ideal, the
maximum radial velocity is indepen-
dent of the gas density. The results for
any case are therefore insensitive to
the nebular mass.
The particle properties and drag law
determine the velocity components cor-
responding to a given particle size. The
150
CARNEGIE INSTITUTION
Upper Limit (%-V.)
t e «t
Radial Drift
Perturbed Kepler
Orbit
particle radius
Fig. 23. Radial velocity as a function of particle size (schematic). The shape of the curve depends
on the drag laws, but the value of the upper limit does not.
solution is complicated by the fact that
any of four drag laws may be appli-
cable, depending on the Reynolds
number and the ratio of mean free path
to particle size. Each drag law must be
tried until a self-consistent solution is
found. I have written a computer pro-
gram that performs this task, solving
the equations of motion numerically.
The velocity components can be deter-
mined for a (spherical) particle of any
size and density in a nebula of arbi-
trary structure. Results to date must be
considered preliminary, but the follow-
ing conclusions appear to be justified
for a wide range of "reasonable" nebu-
lar models:
(1) The maximum radial velocity
can be large. In the inner nebula, it
may exceed 100 m/s ] at 1AU, decreas-
ing by about an order of magnitude in
the outer part. (2) These velocities gen-
erally correspond to particle of radii ~
10-100 cm (Fig. 24). (3) Characteris-
tic lifetimes for spiraling into the sun
can be very short for such particles,
ranging from a few tens of years in the
inner nebula to 10 4 years in the outer
part. The loss of solids from any region,
or mixing between regions, could have
been significant, depending on the par-
ticle size distribution. (4) For small
particles (t e «t p ), larger and/or den-
ser bodies are removed more rapidly.
For large particles (t e »t p ), the oppo-
site is true. Fractionation by composi-
tion, e.g., Fe from silicates, could occur
in either direction. The rate of mass
loss is greatest in the inner nebula;
Mercury's high Fe content may be ex-
plainable by this effect. (5) Radial
and transverse velocities are strongly
dependent on particle size. It is gener-
ally assumed that gas drag would
damp their relative motions. However,
if the particles vary in size, they may
have large relative velocities because
of the gas. It is not clear whether the
growth of large bodies would be aided
or hindered. High relative velocities
would inhibit gravitational instability
(Goldreich and Ward, 1973), but if par-
ticles stick together on impact, the in-
creased rate of collisions might be con-
ducive to growth.
These and many other consequences
remain to be investigated in greater
detail.
DEPARTMENT OF TERRESTRIAL MAGNETISM
151
I0 :
10 -
10-
-[ — I I I I 1 1 1 1 1 — I I II 1 1
Peer -3
T = (600/r)°K
P„ = 3 g cm-3
r(AU)
Fig. 24. Radial velocities of particles of vari-
ous radii (S) as functions of location in a typical
model nebula. Particle density = 3 g cm -1 . Nebu-
lar gas pressure varies as r~\ density as r~ 2 .
References
Goldreich, P., and Ward, W. R., The forma-
tion of pbanetesimals, Astrophys. J. 183,
1051, 1973.
Whipple, F. L., Radial pressure in the solar
nebula as affecting the motions of
planetesimals, NASA SP-319, 355,
1973.
Interplanetary Bodies and
Solar System History
George W. Wetherill
Studies of the solar system and its
origin have been primarily directed
toward the largest bodies — the sun and
the planets. However, during the past
few decades it has become increasingly
clear that equally significant informa-
tion can be obtained from the smaller
bodies of the solar system: the satel-
lites of the planets including our moon,
and the asteroids, comets, meteoroids,
and meteorites. The small size of these
bodies has permitted their record of
primordial events in solar system his-
tory to be preserved to a much greater
extent than is the case for hot and in-
ternally active planets. The inter-
planetary orbits of comets and many
asteroids produce a cratering record on
the moon and planets which reveals
their size and number throughout the
history of the solar system. Study of
meteorites and lunar samples permits
direct chemical investigation of the
record of events dating back to the for-
mative period of the solar system. Re-
cent evidence for presolar grains in
meteorites (Clayton and Mayeda, 1975)
provides a link between planetary for-
mation and the presolar galactic his-
tory of the matter from which it was
formed.
Interpretation of the data obtained
from studies of these small bodies re-
quires understanding of their sources
in the solar system and the evolution of
their orbits. Very significant advances
in these areas have followed the devel-
opment of new techniques for tracing
their orbital evolution over billions of
years (Opik 1951, Arnold 1965), for
understanding the detailed structure of
the asteroidal regions (Williams 1969,
1971), and for calculation of the devel-
opment of the swarm of interplanetary
bodies which preceded the formation of
the planets themselves (Safronov
1969). Examples of application of these
achievements are given in the follow-
ing three subsections, which deal with
the role of interplanetary bodies in dis-
tinct periods in solar system history:
(1) the formation of the planets and the
moon (4.5 billion years ago); (2) the late
heavy bombardment of the moon and
terrestrial planets which produced the
heavily cratered terrains preserved on
the lunar highlands, Mercury, and
Mars (4.0 billion years ago); and (3) the
earth-crossing Apollo objects that have
dominated the cratering record of the
earth and moon during the past three
billion years.
152
CARNEGIE INSTITUTION
The Role of Large Impacting Bodies in
the Formation of the Earth and Moon
Introduction. Over the years many
processes have been suggested which
involve the existence of large (> 10 24 g,
~ 500 km radius) planetesimals during
the formation of the solar system.
These include lunar-sized meteorite
parent bodies (Urey, 1959), capture
theories for the formation of the moon
(Gerstenkorn, 1969; Singer, 1972), dis-
integrative capture by passage of
bodies within the earth's Roche limit
(Opik, 1972; Smith, 1974; Mitler,
1975), impact heating of the earth (Saf-
ronov, 1969) and moon (Ruskol, 1974),
tilting of the earth's rotation axis by a
small number of large impacts (Saf-
ronov, 1960), formation of the moon by
"splashes" resulting from grazing im-
pacts with the earth (Hartmann and
Davis, 1975; Cameron and Ward,
1976), predifferentiation of lunar
planetesimals (Kaula and Bigeleisen,
1975), and the existence of large post-
accretional residual bodies needed to
produce the late heavy bombardment
which continued for —600 m.y. after
the end of the accretion of the planets
(Wetherill, 1975b). An evaluation of
the plausibility and relative impor-
tance of these processes requires a
theory that predicts at least semiquan-
titatively both the size and the velocity
distribution of the planetesimals that
accumulated to form the terrestrial
planets and the moon. Considerable
progress toward such a theory has been
made by Safronov (1969) and his co-
workers (Zvyagina et al., 1973). Other,
less comprehensive treatments of this
problem have been given by Marcus
(1965, 1969), Hallam and Marcus
(1974), Hartmann and Davis (1975),
and Weidenschilling (1974). A charac-
teristic of these theories is that ac-
cumulation of the planetesimals into
the terrestrial planets and the moon by
random encounters is slow (10 5 to 10*
years) compared with the time scale of
~10 :i yr proposed for accumulation
from a gaseous solar nebula rather
than from a protoplanetary swarm of
planetesimals (Cameron, 1973). Seri-
ous difficulties with the latter type of
theory have been discussed (Safronov,
1975). In spite of these difficulties,
gravitational accretional heating as-
sociated with very rapid accumulation
has been the starting point of most dis-
cussions of the geochemical and
geophysical evolution of the moon, and
it is commonly stated that such dynam-
ically unlikely rapid accretion is re-
quired if one is to understand the evi-
dence for early chemical differentiation
of the moon.
As discussed by Safronov (1969), ap-
preciable accretional heating does not
require rapid accretion, provided that a
sufficiently large number of the ac-
cumulating bodies are massive and
energetic enough to deposit a signifi-
cant quantity of energy deep beneath
the accretional surface and thereby to
preclude loss by surface radiation into
space.
Calculation of the size distribution of
the largest bodies is not readily amen-
able to coagulation theory (Safronov
and Zvyagina, 1969) even though con-
siderable insight is provided by this
approach to the size distribution of
smaller bodies, especially when the
combined effects of accumulation and
fragmentation are considered
(Zvyagina et al., 1973). One reason for
the failure of the theory for the largest
accumulating bodies is that for these
bodies the gravitational field of the
growing body can augment its effective
cross section in such a way that the
cross section becomes nearly propor-
tional to the fourth power of its radius
rather than to the geometrically de-
termined square of the radius. This ef-
fect is partially included in the only so-
lutions to the coagulation equation
which have been reported, i.e., for a
coagulation kernel linear in the mass
and hence proportional to the cube of
the radius. This approximates the situ-
ation over much of the mass range but
not at the uppermost end of the range,
in which a small difference in radius
might, at least in principle, lead to a
great difference in the rate at which
DEPARTMENT OF TERRESTRIAL MAGNETISM
153
the radius increases. Numerical exam-
ples of such "runaway accretion" have
been given by Hartmann and Davis
(1975).
Safronov (1969) has made use of
runaway accretion to predict a "second
largest mass" in the earth's accumula-
tion zone of 10~ ! earth masses, i.e., 6 x
10 24 g. In the present work a variant of
this model is proposed which leads to
the second largest body being more
massive, and thereby also implies sig-
nificantly greater accretional heating
of the moon.
Runaway accretion. The rate of
growth of the ratio of the radii of the
two largest bodies (R = rjr. z ) is by the
elementary rules of differentiation:
dR
dt
L d
dr x
df
R
dr,
dt
(1)
In a simple accumulation model in
which a body of radius r sweeps up
smaller bodies with relative velocity V,
dr_
dt
p,.(t)
V
(i + Jfc)
(2)
where p c (t) = the spatial density of
mass in the form of smaller bodies, p„ =
the density of the larger body, and Ve
= the escape velocity of the larger
body.
In Safronov's accretion theory the
mean relative velocity is related to the
escape velocity of the largest body by
use of the dimensionless ratio 6:
V 1 =
2H ~
and
=
GM„
R,y 2
(4)
where G is the gravitational constant,
and the subscript p refers to the escape
velocity, mass, and radius of the
largest body in the earth's zone. Saf-
ronov, following earlier work of
Gurevich and Lebedinskii (1950), pre-
sents calculations that support his con-
clusion that 6 varies slowly (i.e., by a
factor of ~2), while M p increases by a
factor of ~10 12 during accretion. This
implies that the mean relative velocity
"keeps step" with the growth in radius
and escape velocity during accumula-
tion. The numerical value of 6 depends
on the balance between acceleration by
mutual perturbations by planetesimals
and dissipation of energy in collision of
planetesimals. Under a variety of con-
ditions in which accretion occurs after
loss of most of the nebular gas (appro-
priate to earth's zone), the value of 6 is
estimated to range from about 7 down
to about 1. Therefore, the mean veloc-
ity with which smaller bodies collide
with the largest body is always less
than its escape velocity. Calculations
such as those reported by Marcus
(1969) using experimental data of
Gault et at. (1963) then lead to the con-
clusion that at these velocities the
largest bodies will grow in radius
rather than fragment. Smaller bodies
may either accrete or erode, depending
upon their mass.
Substitution of 6 into (2) permits (1)
to be developed to:
dR
dt
\/ p c (t)\ (2d-R)(R-l)\
(5)
where G is the gravitational constant.
For constant 6, the solution of (5)
may be written as:
1 +28
(3) R
L 20-R J v 26 + 1 y
I- 2Q-R... J \ 9,ft+-[ /
(6)
where it! (l is the initial value of/?,
A- (20+1) Pr (o)\ I- —1 V2
L 3 6p„ -I
and f is a time-like variable:
f i: =—^~ f p ( .(t)dt.
p r (o) J " H
154
CARNEGIE INSTITUTION
Equation (6) shows that the extent of
the runaway accretion is limited, as
discussed earlier by Safronov (1969).
As t becomes indefinitely large, R—>26,
which is therefore the maximum ratio
of the final radii. For nonzero values of
the initial radius
R<26 (7)
The final ratio of the masses of the two
largest bodies is therefore less than
(20) 3 . By use of 6 = 5 and assuming t*
to be infinite, Safronov obtained the re-
sult mentioned earlier, that the second
largest body in the earth's zone had a
mass of ~ 6 x 10 24 g.
As will be seen later, in this case the
number of bodies surviving to the ter-
minal stages of growth of the moon will
be too small to produce much accre-
tional heating.
However, there is no reason to sup-
pose that 6 must be as large as 5. In
fact, Safronov (1969, Table 8, Chapter
7) gives 6—2 during the stage of
planetesimal growth prior to the time
in which most of the mass is in a single
largest body, the planetary embryo.
This value limits the mass ratio to less
than 64. It is very difficult to know the
correct value of 6 to be used at this
stage, as direct knowledge of the ap-
propriate dissipation to be assumed in
protoplanetary planetesimal collisions
is not available. However, the values
assumed by Safronov to obtain 6 — 2
appear plausible. During later stages
of growth of the planetary embryo, Saf-
ronov shows that the value of 6 in-
creases to ~5, implicitly assuming that
the remaining planetesimals cross the
orbit of only one dominant planetary
embryo, moving in a circular orbit. In-
sofar as this assumption is correct, this
is a consequence of the fact that succes-
sive encounters with the embryo will
always occur at the same value of the
mean geocentric velocity, provided that
no perturbations by a third body occur
between these successive encounters.
This in turn is a consequence of the
conservation of energy in these en-
counters in the geocentric reference
frame, related to the constancy of the
Tisserand criterion and the Jacobi in-
tegral in the restricted three-body
problem.
The fact that when the earth has
reached —10% of its final mass, bodies
with geocentric velocities of 2.5 km/s
will cross the orbit of Venus invali-
dates the above assumption. The case
where a small body is crossing the orbit
of both Earth and Venus has been in-
vestigated both by analytical and
Monte Carlo techniques, and it has
been found that Venus-crossing results
in 6 decreasing to the range 2 to 0.5
during the final stages of planetary
accretion, rather than increasing to
~5, as is obtained when this effect is
not considered. This leads to mean ve-
locities of ~8 km/s for objects crossing
the orbits of both Earth and Venus. Re-
sults of calculations of the masses of
the second largest body are given in
Table 2.
Elimination of the residual planetes-
imals. These lower values of 6 lead to
much larger objects in the zone of
Earth and Venus during the final
stages of accumulation. In the case of
runaway accretion the overwhelming
dominance of the planetary embryo
was established by the much more
rapid growth of the largest body. For
these lower values of 6, this dominance
is established by an alternative effect:
the greater stability of the embryo dur-
ing the regime of fragmentation which
exists during the final stage of accumu-
lation of the embryo. This fragmenta-
tion primarily results from two pro-
cesses: disruption following approach
within the Roche limit of the Earth or
Venus, or collisional fragmentation.
The largest residual planetesimals
(~10 2,i g) will probably be immune to
destruction by collision and are likely
to be continuing to accrete until they
are disrupted by tidal forces. Their
mass is greater than that of the present
moon, which is near equilibrium with
respect to accretion vs. erosion by crat-
ering (Gault et al., 1963; O'Keefe and
Ahrens, 1975, 1976) at an average im-
pact velocity about twice as great. At a
DEPARTMENT OF TERRESTRIAL MAGNETISM
155
TABLE 2.
A. Mass of the Second Largest Body When the Earth Embryo is 2
Initial Mass Ratio atM, = 10 21 g:
x 10 27 g
6
1
2
3
4
5
1.16
1.1 x 10 27 g
4.0 x 10 2,i
1.9 x 10 2,i
1.1 x 10 2,i
7.2 x lO 25
2.20
4.2 x 10 2li g
7.4 x 10 25
2.4 x 10 25
1.1 X lO 25
6.4 x 10 24
8.00
2.0 x 10 2,i g
3.9 x 10 25
1.2 x 10 25
5.5 x 10 2J
2.9 x 10 2J
B. Mass of the Second Largest Body When the Largest Body is 4
x 10 27 g
1
2
3
4
5
2.2 x 10 27 g
7.0 x 10 2B
3.0 x 10 2,i
1.6 x 10 2 «
1.1 x 10 2li
6.4 x 10 2,, g
1.3 x 10 2,i
4.3 x lO 2 '
2.0 x 10 25
1.1 x 10 25
5.0 x 10 2,i g
7.7 x 10 25
2.3 x 10 25
1.0 x 10 25
5.5 x 10 2J
mean impact velocity of 8 km/s, bodies
less than ~ 2 x 10 24 g in mass will be
subject to fragmentation rather than
continuing to accrete. The lifetime for
collisional destruction of these bodies
by smaller residual bodies in the zone
of Earth and Venus may be estimated
by methods used earlier (Wetherill,
1967) to be in the range of 10" to 10 7
years. Smaller bodies of mass less than
10 22 g will be very short lived (<10 5
yr), and a collisional mass distribution
may be expected to be established
among these smaller bodies. The popu-
lation of small bodies will be main-
tained by supply of ~10 2i g bodies re-
sulting from Roche limit destruction of
bodies <10 24 g in mass. Fragments will
be removed from this population of
fragmental material by accretion onto
those bodies large enough to accrete
(i.e., < 2 x 10 24 g), which will ulti-
mately be only the Earth and Venus
embryos. Therefore, the final result
will be the same as was the case for the
alternative of runaway accretion. All
the mass will be in the planetary em-
bryos (except for a small fraction accel-
erated to Jupiter-crossing). The dif-
ference will be that large bodies, up to
~3 x 10 2,i g in mass will have had a
transient existence during the final
stages of accretion. The time scale for
sweep-up of all the bodies will be
somewhat extended by the higher ve-
locities during the final stages but will
still be ~10 8 yr.
Heating of the moon by impact of
large planetesimals. As discussed ear-
lier (Safronov 1969, Ruskol 1974,
Wetherill 1975a), large high-velocity
impacts will deposit their kinetic en-
ergy beneath the lunar surface, which
will be buried by fallback of crater
ejecta into the transient cavity. This
trapping of energy will be particularly
pronounced for the largest impacts in-
asmuch as in this case most of the
fragmented material will fail to escape
the cavity because of the large gravita-
tional energy required.
In an accreting body, the heat trap-
ped in this manner will be further
buried by deposition of newly accreted
material onto the floor of the crater,
thereby minimizing the radiation into
space which limits the temperature in
most accretion models (i.e., Hanks and
Anderson, 1969; Mizutani et al., 1972).
This effect has been estimated by a
simple model in which the depth of bu-
rial of the heat increases as a result of
accretion and decreases by conduction
of heat to the surface.
Under the conditions of accretion,
the dominant mode of heat transfer to
the surface will not be ordinary lattice
conduction but mixing of previously
buried material to the surface by sub-
sequent impacts. The appropriate
156
CARNEGIE INSTITUTION
value of "eddy conductivity" to be used
is quite uncertain. Values of —2 cm 2 /s
are calculated by Safronov (1969), and
similar values can be estimated from
the work of Blake and Wasserburg
(1976) carried out in a different con-
text. Use of this value indicates that
bodies with radii larger than —40 km
in radius, i.e., with masses <10 21 g will
have a large fraction of their kinetic
energy permanently trapped beneath
the accreting surface. At first the frac-
tion of impact melt produced in any
single impact will not be large. Much of
the melted material will be mixed with
pulverized unmelted crater debris; it
will merely heat it slightly and produce
little igneous differentiation. However,
the cumulative effects of further im-
pacts will cause the temperature of the
multiply reworked debris to approach
and exceed the melting point. Very
close estimates are difficult, but pre-
liminary calculations suggest that dur-
ing the addition of the final half of the
moon's mass (outer 360 km) about 10%
of the impacting mass was in the form
of bodies more massive than 10 2 " g.
This leads to an average kinetic energy
trapped within the outer 360 km of ~8
x 10 9 ergs/g. This is about 40% of the
energy required to produce complete
melting of this region. If these figures
are correct, the outermost portion of
the moon will be heterogeneously
melted and differentiated at the end of
accretion. The individual large accret-
ing bodies will be sufficiently energetic
to completely melt many times their
own volume of material, which has al-
ready been heated much of the way to-
ward melting point. Regions which by
chance failed to be impacted by a very
large body would contain only small
regions of total melt and much fine ma-
terial mixed with quenched glass. The
resulting state of primordial dif-
ferentiation may be quite different
from a "magma ocean" resulting from
uniformly distributed heat sources.
In the alternative model, in which
runaway accretion causes the second
largest body to be much smaller, it
seems unlikely that lunar heating
would be sufficient to account for the
evidence for primordial lunar dif-
ferentiation, especially if the larger
bodies are still destroyed by encounters
within the Roche limit. The resulting
trapped energy in the outer 360 km is
estimated to be ~ 3 x 10 7 ergs/g, suffi-
cient to melt only —0.1% of this region.
Because of many uncertainties in these
calculations, the absolute values of
these numbers should probably not be
taken too seriously. However in the al-
ternative model, the much smaller
number of large bodies, impacting at
less than half the velocity, must in-
evitably result in much smaller quan-
tities of trapped energy.
Concluding remarks. This rather
sketchy account of a chain of events
within the framework of the Safronov
accretion theory should at least show
that significant primordial differentia-
tion of the moon could have occurred
even on a long — 10 8 time scale. The
subsequent magmatic evolution of a
heterogeneously melted and dif-
ferentiated moon will be more complex
than that of the "magma ocean." Al-
though this complexity may be an in-
convenience, in no way does complexity
or simplicity bear on the question of
whether the moon was uniformly or
heterogeneously differentiated. The
possibility of heterogeneous dif-
ferentiation was not introduced as an
ad hoc complexity to complicate the
lives of petrologists; it is merely a
probable consequence of an equally
simple, and dynamically more plausi-
ble, model of accretion. It may be that
exploration of the petrological conse-
quences of this variant may result in
its seeming more satisfactory than the
magma ocean, or vice versa. This re-
mains to be done; some speculations
bearing on this have been presented
(Wetherill 1975a).
Primordial heating of the terrestrial
planets would also result from this
mechanism. The differences in the ini-
tial temperatures between the larger
bodies such as the earth and the moon
would be less than in the case of heat-
ing by the very rapid accretional mod-
DEPARTMENT OF TERRESTRIAL MAGNETISM
157
els that have been published (e.g.,
Hanks and Anderson, 1969). This
arises from the fact that in the theory
presented in this work the energy
input depends primarily upon the kine-
tic energy of the heliocentric planetes-
imals, the acceleration by the gravita-
tional field of the accreting planet
being of secondary importance. How-
ever, somewhat higher initial tempera-
tures may be expected for the earth, if
only as a consequence of all the impacts
being caused by heliocentric bodies
rather than largely geocentric bodies,
as is the case for a coaccreting moon.
Insofar as planetesimals in the range of
10 25 - 10 2,i g are. still accreting prior to
tidal disruption, significant heating
and even igneous differentiation of
their interiors are possible but difficult
to estimate with any degree of cer-
tainty.
As mentioned in the introduction,
numerous other authors have invoked
large bodies in connection with various
problems of lunar origin. The accre-
tional model outlined herein can, when
more fully worked out, lead to fairly
definite predictions concerning the size
(distribution, orbits, and evolution of
large bodies in the early solar system.
In this context these suggestions by
others represent processes which are
possible, rather than being contrived,
I as has sometimes been alleged. It re-
mains to be seen to what extent they
are probable rather than simply possi-
ble.
It is also of interest to inquire into
the ultimate fate of the large bodies
that were initially present in the inner
solar system. Most of them simply ac-
creted to form the moon and terrestrial
planets. A small percentage of them
were ejected from the solar system.
However, the Monte Carlo calculations
mentioned earlier indicate that if Mars
was present during the accretion of the
earth, about 1% of the earth planetes-
imals remaining after the growth of
the earth was essentially complete
would be removed from earth-crossing
by Mars perturbations. Most of these
would have been deep Mars-crossers in
orbits similar to the present Amor ob-
jects and returned to earth-crossing on
a time scale of ~2 x 10* years. The role
of these bodies in the late heavy bom-
bardment of the terrestrial planets has
been discussed elsewhere (Wetherill
1975b). Of greater interest are the
smaller number ( — 10%) of these which
evolve into shallow Mars-crossers with
lifetimes comparable to the age of the
solar system. Of the ~10 2li g which re-
mained when the growth of the earth
was 98% complete, perhaps 10 2i g was
"implanted" into these long-lived or-
bits. If Mars accumulated on the same
time scale as the earth, those bodies
would be outnumbered by similar re-
sidua from the formation of Mars. On
the other hand, the formation of Mars
is difficult to understand unless it rep-
resents an "aborted" earth-size planet,
the growth of which was terminated by
the early heavy bombardment of Jupi-
ter planetesimals, as suggested by Saf-
ronov (1969) and Weidenschilling
(1975). In this case the Mars residua
would have been destroyed before the
implantation of the earth planetesi-
mals, and this region of the asteroid
belt would now be largely populated
with residual material from the forma-
tion of Earth and Venus. The dif-
ferentiated meteorites, such as the
basaltic achondrites, are likely to arise
from this region of the asteroid belt,
and it is possible that these objects are
fragments of earth planetesimals that
have now returned home after an ab-
sence of 4.5 billion years. When viewed
from this standpoint it is possible that
the dynamic and chemical heterogene-
ity of the present asteroid belt is a con-
sequence of this region of the solar sys-
tem being a "sink" for residual mate-
rial from both the inner and the outer
solar system.
Late (4.0 b.y.) Bombardment of the
Moon and Terrestrial Planets
Introduction. Prior to the recent pro-
gram of manned and unmanned explo-
ration of the moon and terrestrial
planets, there was a tendency to sepa-
158
CARNEGIE INSTITUTION
rate earth history into two discrete and
nonoverlapping periods of time. These
were the last ~ 3.5 x 10 9 yr, revealed
by the terrestrial geological record, and
the primordial nebular and accretional
periods, 4.6 x 10 9 years ago. Evidence
concerning the latter was inferred from
theoretical considerations and observa-
tional and meteoritic data. It was
commonly believed that the cratered
surface of the moon represented bom-
bardment during the terminal phase of
this primordial period.
Following the Apollo 14 mission to
the Fra Mauro region of the moon and
dating of these rocks at 3.9-4.0 x 10 9
years, it became apparent that the
period of intensive lunar cratering and
mare basin formation continued long
after the completion of the — 10 8 -year
formational period of the moon and
planets. In addition to the "early heavy
bombardment" which more or less ac-
companied the accretion of the planets,
there was a "late heavy bombardment"
which continued for hundreds of mil-
lions of years and may well have been
episodic in character (Tera etal., 1974).
The Mariner 10 spacecraft photo-
graphed —40% of the surface of Mer-
cury to a resolution of 10 km and
selected regions to a resolution of 200
m. Photogeological study of these data
indicates a remarkable similarity be-
tween the impact flux on Mercury and
the moon (Murray et al., 1975). Both
planets contain densely cratered re-
gions containing more than 200
craters/10" km 2 larger than 10 km in
diameter. The lunar regions of this
kind represent a steady-state crater
population, whereas Mercury contains
both steady-state regions of this kind
and poorly understood "intercrater"
terrains in which a high density of
craters larger than 50 km in diameter
appears to be superimposed on an older
relatively uncratered surface (Trask
and Guest 1975). Both planets contain
ringed impact basins and associated
ejecta deposits ranging up to ~ 1000 km
in diameter. In both cases postbasin
surfaces have crater densities of —120
craters/10" km 2 greater than 10 km
diameter. Finally, both planets contain
younger basin-filling surfaces presum-
ably of volcanic origin with lower cra-
ter frequencies of —40 craters/10" km 2
greater than 10 km in diameter. Al-
though the Martian crater record is
obscured by erosional and depositional
processes, photogeological studies
suggest a similar impact history for
that planet (Wilhelms, 1973; Chap-
man, 1974; Soderblome^a/., 1974).
These similarities between the three
planetary bodies could be coincidental
or could be the result of their experi-
encing a very similar impact flux his-
tory. If the latter is the case, the rela-
tively lightly cratered postbasin sur-
faces with crater counts of —120/10"
km 2 represent a "marker horizon"
throughout the inner solar system cor-
responding to a time of 3.9±0.1 b.y.
ago, as indicated by the lunar
radiometric chronology.
The purpose of this work is to quan-
titatively explore alternative models
for the interplanetary flux during the
first — 10 9 years of solar system history
in order to evaluate whether any of
these are compatible with the hypothe-
sis of a uniform flux history for these
planets. The other consequences of
these models must also be examined in
order to understand the extent to
which they can be reconciled with
other features of the solar system, its
origin, and history.
Size distribution of the impacting
bodies. Insofar as the crater density is
not so high as to represent a steady
state between crater production and
destruction (Gault 1970), the size dis-
tribution of craters can be used to infer
the size distribution of the impacting
bodies. Although the most densely cra-
tered regions of the lunar highlands
are in such a steady state, the more
lightly cratered plains regions such as
the Fra Mauro and Cayley plains are
well below a steady-state density.
Detailed investigations of lunar cra-
ter size distributions (Neukum et al.,
1975) show that the pre-Imbrium,
plains, and mare surfaces contain the
same size distribution of craters. This
DEPARTMENT OF TERRESTRIAL MAGNETISM
159
distribution cannot be described by a
single-power law exponent over the en-
tire range of crater sizes. For the
largest craters (>10 km diameter) the
slope of the distribution is rather flat
and approximately corresponds to a
cumulative distribution:
N = N D ~ a
where a is in the range 1.4 to 1.6. This
is in agreement with the results of pre-
vious authors (Shoemaker et al., 1962;
Baldwin, 1963) and represents a pro-
jectile mass distribution in which the
total mass is strongly concentrated in
the largest projectiles.
If the size range of bombarding pro-
jectiles is continued to include the Im-
brium planetesimal, this concentration
of mass in the largest bodies is at least
equally pronounced. The projectile
mass required to produce the large
mass craters can be estimated by com-
bining mare crater counts (Shoemaker
et al., 1962) with a reasonable energy-
scaling law based on nuclear explo-
sions and which relates diameter to
impacting mass:
D = 0.115 M m
where D is the crater diameter in me-
ters, M is the impacting mass in grams,
and an impact velocity of 15 km/s is
used. This calculation gives 3 x 10
as the total flux since the filling of the
mare basins. The integral flux on the
Fra Mauro and Cayley plains forma-
tions is about three times as great, cor-
responding to a mass flux of ~ 10 20 g. A
similar calculation for the mass of the
Imbrium planetesimal gives a mass of
~10 21 g. In the same way, the mass of
the Orientale projectile comes out to be
—2 x 10 2<) g. Therefore, up to and in-
cluding projectiles of Imbrium size
(—50 km in radius), the large pro-
jectiles are characterized by the fact
that the largest body has a mass
greater than that of the remaining ob-
jects.
This is not the kind of size distribu-
tion expected for objects that are in a
steady state resulting from mutual col-
lisional comminution. It more closely
resembles the end product of an accre-
tional process in which larger bodies
have grown at the expense of smaller
ones. In the treatment of specific mod-
els for the sources of the late heavy
bombardment, one relevant considera-
tion should be whether or not a colli-
sional mass spectrum should be ex-
pected under the conditions assumed
for the model.
Insofar as there is some propor-
tionality between the quantity of im-
pact breccias and the extent of impact
resetting of measured ages, a mass dis-
tribution dominated by a few large im-
pacts will produce an episodic age dis-
tribution. Therefore the ~ 30-fold de-
crease in mass flux between the time of
the Imbrium impact and the older
mare surfaces need not be entirely ex-
plained in terms of a uniformly decay-
ing mass flux but will be in a major
way a consequence of a statistical fluc-
tuation in this flux. Therefore, to a cer-
tain extent episodicity will result sim-
ply from the "lumpiness" of the size
distribution. This effect in itself is in-
sufficient to produce simultaneous
episodes on several planets. However,
approximate simultaneity of the cessa-
tion of the late heavy bombardment
could result from a flux decreasing
with a characteristic half-life of —50
g m.y.
Relative bombardment rates on the
moon and terrestrial planets. The
hypothesis that the density time scale
of the lunar craters can be extended to
the other terrestrial planets requires
that there have existed one or more
populations of bodies in the inner solar
system which had a nearly equal prob-
ability of impacting a given area on
any of the planets. Candidate popula-
tions will be characterized by their or-
bital element and mass distributions
and by their total mass.
The two principal classes of orbits
which should be considered are those of
bodies in heliocentric orbit and those
orbiting as satellites of a terrestrial
planet. Apart from the possibility that
a constant crater density may arise by
coincidence, the latter class must
160
CARNEGIE INSTITUTION
evolve into heliocentric orbits in order
to impact planets other than the one
they are initially orbiting. No detailed
discussion of this evolution from
planetocentric to heliocentric orbits
exists, although treatments of tidal
evolution of satellites (reviewed by
Kaula, 1971) and geocentric swarms
(Ruskol, 1960, 1963, 1972; Ruskol et
al., 1975; Kaula and Harris, 1975) are
relevant to this problem. This class of
initial orbits will not be discussed here.
However, this possibility should not be
forgotten.
The evolution of various classes of
heliocentric interplanetary orbits has
been discussed previously (Wetherill
1974). This treatment included a dis-
cussion of the modifications made in
the Monte Carlo techniques which
permitted an adequate statistical sam-
ple of impacts on small bodies such as
the moon and Mercury. The results of
these calculations are given in Table 3.
The first two entries in this table
represent bodies evolving from initial
orbits similar to those required to ex-
plain the orbital distribution of the
Prairie Network fireballs, and the ob-
served time of fall and radiants of
chondritic meteorites (Wetherill 1968,
1971). Such orbits can evolve from
those of short-period comets. The sec-
ond entry in Table 3 is such an object,
Comet Temple 2. These data are nor-
malized so that the lunar impact prob-
ability per unit area is assigned the
value of unity. The impact probability
per unit area on the four terrestrial
planets is seen to agree with the lunar
value within a factor of about 3. This
result is largely independent of the
large difference in the efficiency with
which these bodies impact the planets,
shown in the third column. Objects ini-
tially in orbits similar to those of the
first entry will very frequently strike
an inner planet; 24% of them will hit
the earth. In contrast, those in orbits
similar to that of the second entry will
more often be perturbed into Jupiter-
crossing and ejected by that planet into
interstellar orbits. Only a fraction of a
percent of these bodies will impact a
terrestrial planet.
The next three entries represent ini-
tial orbits that penetrate more deeply
into the inner solar system, penetrat-
ing the perihelia of Venus or Mercury.
The orbits calculated are those of ac-
tual bodies in the present solar system,
but this is done only for illustrative
purposes and not to suggest that these
modern objects have any relation to the
late heavy bombardment. The objects
with the smaller perihelia will cause a
greater crater density on Mercury than
on the moon and Mars. Strictly speak-
ing, these initial orbits are not consis-
tent with a uniform flux history. How-
TABLE 3. Relative Impact Probability/Unit Area*
Initial Orbit
Earth
Impacts
Aphelion
Perihelion
(%)
Mercury
Venus
Earth
Mars
1
1.01
4.00
24.0
0.30
0.94
2.26
1.15
2
1.36
4.67
0.11
0.69
1.26
1.78
0.71
3
0.34
4.18
6.0
4.21
2.37
1.23
0.37
4
0.19
1.97
18.5
7.50
1.97
1.08
0.45
5
0.70
3.61
5.2
1.60
2.34
1.34
0.27
6
1.27
1.89
27.8
1.08
2.58
2.82
11.9
7
0.98
0.99
45.0
1.05
3.02
3.99
0.96
8
0.67
0.82
29.6
1.08
2.44
1.36
0.18
9
0.33
0.41
<0.01
>1.5 x 10 4
* Normalized to Moon = 1.0.
DEPARTMENT OF TERRESTRIAL MAGNETISM
161
ever, an admixture of such orbits with
others with perihelia nearer to that of
the earth (the first two entries) would
lead to an approximately uniform flux
history as an average result.
The last four entries represent initial
orbits of sufficiently low eccentricity to
cross the orbit of only one planet. It
might be thought that this would
greatly favor impacts on that planet.
To some extent this is true. The Mars-
crossers are about ten times more
likely to strike Mars (per unit area)
than the moon. Mercury-crossers only
very rarely evolve into even Venus-
crossing before they strike Mercury.
The results for these two planets are
largely a consequence of these small
planets having insufficient mass to
readily "pump" the eccentricity to high
enough values to cross another planet-
ary orbit. In contrast, the larger ter-
restrial planets, Venus and the earth,
perturb bodies into highly eccentric or-
bits that have a good chance of striking
the moon or any of the terrestrial
planets.
The general result of these calcula-
tions is that planetary perturbations
tend to spread interplanetary bodies
throughout the inner solar system, re-
sulting in a comparable flux on the
moon and terrestrial planets. There are
exceptions to this; those which initially
cross only Mercury are a particularly
striking exception. Nevertheless, it
will usually be difficult to confine an
interplanetary population primarily to
a particular planet, and many distribu-
tions of initial orbits will lead to a simi-
lar flux on the moon and Mercury.
To compare crater densities on the
various planets it is necessary to con-
sider the impact velocity as well. These
have been calculated by the same tech-
niques used to obtain the data of Table
3. Examples of these calculated veloc-
ity distributions are given in Figs. 25
and 26. Those orbits with initial
perihelia well beyond Mercury will im-
pact Mercury with about two to three
times the velocity they impact the
moon. When this result is combined
with the size distribution discussed in
Initial Orbit: (Mars Grosser)
Aphelion = 1.90 A.U.
Perihelion =1.27 A.U.
"
—
Inclination = 6°
-
Mercury
\—V— r— ■ , ,
Moon
Lti
10 20 30 40
RELATIVE VELOCITY
50
60 ?0
KM/SEC
Fig. 25. . Distribution of impact velocities on
the moon and Mercury for a swarm of bodies ini-
tially in Mars-crossing orbit. The velocity is
given as V a , i.e., prior to the additional accelera-
tion resulting from the gravitational field of the
planet. For small planets like these this effect is
small.
the previous section, it turns out that
the effect of velocity is to increase the
crater density by a factor of about 2.5.
For objects such as those given in the
first two entries of Table 2, this effect
offsets the factor of 2 to 3 lower impact
probability on Mercury relative to the
moon, which arises from the rather in-
effectual "Earth and Venus barrier"
these bodies must surmount. This ve-
locity effect is less pronounced for or-
bits that are initially Mercury-
crossing, as shown in Fig. 26. These
bodies produce a greater flux on Mer-
cury than on the moon. The effect of
velocity is thus seen to partially offset
the effect of the initial perihelia, tend-
ing to result in more uniform crater
densities.
Storage places. Much of the earlier
discussion of this problem has centered
on the question of whether there are
places in the solar system in which
large bodies can be "stored" for hun-
dreds of millions of years prior to their
impacting the moon or planets. Typical
lifetimes, given by Monte Carlo calcu-
lations, characterizing the stability of
various types of initial orbits, are given
in Table 4. It may be seen that two re-
gions of the solar system will store
162
CARNEGIE INSTITUTION
150
50
Initial Orbit: (Comet Encke)
Aphelion =4.18 A.U.
Perihelion = 0.34 A.U.
Inclination =D°
Mercury
rr
£ 50
Moon
10 20 30 40 50 60 70 80 90
RELATIVE VELOCITY (V^) KM/SEC
Fig. 26. Distribution of impact velocities for bodies initially in orbits equal to that of periodic
Comet Encke.
bodies for the required length of time.
They are the inner Mars-crossing edge
of the asteroid belt and the outer por-
tion of the solar system in the vicinity
of Uranus and Neptune.
Breakup within the Roche limit.
There are no known storage places
which lead to a simultaneous sharp
peak in the flux on the moon and ter-
restrial planets. Insofar as the photo-
geological and radiometric age data do
indicate a peak flux or "cataclysm" at
— 3.9 b.y. and possibly at other discrete
earlier times, some other explanation
must be found. Several authors have
proposed the breakup of a large body in
the asteroid belt at that time. This is
possible but improbable.
There is another, more probable way
in which ~ 200-km bodies in earth-
crossing orbits may be fragmented. As
discussed by Opik (1951), secular per-
turbation of the argument of perihelion
will cause planet-crossing orbits to
evolve on a time scale of ~10 4 years
into orbits that intersect those of the
crossed planets. Collisions with the
planets can occur during the time
interval for which this intersecting
condition exists. Quantitative evalua-
tion of the problem leads to the well-
known collision formula of Opik that
forms the basis for subsequent Monte
Carlo calculations of orbital evolution.
The same formula can be used to calcu-
late the probability of close encounter
within a given distance of a planet
simply by use of the encounter radius
rather than the planetary radius. The
probability of encounter within a dis-
TABLE 4. Typical Lifetimes of Planet Crossing Orbits
Type of Orbit
Approximate Half-Life
(10" yr)
Earth crosser, aphelion —3 A.U.
Earth crosser, aphelion —4.50 A.U.
Mars crosser, perihelion —1.2 A.U.
Mars crosser, perihelion — 1.5 A.U.
Jupiter crosser
Saturn crosser
Uranus crosser
Neptune crosser
40
10
200
1200
1
2
100
200
DEPARTMENT OF TERRESTRIAL MAGNETISM
163
tance of R e planetary radii is simply
R 2 e times the probability of impact. A
consequence of this is that close
planetary encounters do not in any
sense constitute an improbable or ad-
hoc assumption but are a natural con-
sequence of the same dynamical pro-
cesses that have produced the observed
craters on the moon and planets.
When a small body passes suffi-
ciently close to a planet, differential
gravitational attraction by the planet
will generate a stress field within the
body which, if strong enough, can lead
to its disruption. Roche (1847) consid-
ered this problem for the case of a self-
gravitating liquid satellite and ob-
tained the result:
R.
2.44 ipjp)
1/3
When the encounter distances (in
planetary radii) are less than R e , dis-
ruption will occur; p and p are the den-
sities of the planet and the small body,
respectively. This expression must be
modified in the case of a solid body with
nonzero rigidity. The latter problem
has been treated by Jeffreys (1947),
Opik (1950, 1966), and Sekiguchi
(1970). This earlier work has been re-
viewed and extended by Aggarwal and
Oberbeck (1974). The latter authors
show that a 200-km radius body with a
reasonable tensile strength of 10 7
dynes/cm 2 will undergo tensile failure
along a v complete plane passing
through the center of the body within
an encounter distance of 1.18 (p /p) 1/3
planetary radii. For a "rocky" body
with p = 3 encountering the earth (p ()
= 5.5) this distance is 1.44 planetary
radii, whereas for a predominantly
"icy" body with p = 1, the correspond-
ing encounter distance will be 2.08
planetary radii. Somewhat smaller dis-
tances will be found for Venus encoun-
ters. Because of these assumptions,
very close approaches are required for
disruption. Under these circumstances
the number of disruptive encounters
will be similar to the number of Earth
or Venus impacts. Because of the
greater surface area of Earth or Venus,
impacts with these planets are much
more frequent than lunar or Mercury
impacts (Table 3). Consequently a
large body is more likely to undergo a
disruptive encounter with Earth and
Venus and to distribute fragments to
the moon and Mercury than it is to im-
pact these smaller planetary bodies di-
rectly. For the observed mass distribu-
tion law (equation 8), and disruption
into —100 fragments of more or less
equal mass, lunar impacts of ~10 21 g
fragments of a 10 2:! g body will out-
number direct impacts of 10 21 g bodies
by a factor of about 5. Therefore, even
in this extreme case, disruption can be
expected to produce episodes of heavy
bombardment distributed throughout
the inner solar system.
The discussions given lead to the
conclusions that a modest total mass of
~10 2:i g will suffice to produce the late
heavy bombardment, that storage
places with sufficiently long lifetimes
exist, and that there is at least one
mechanism for producing simultane-
ous episodes of bombardment in the
inner solar system.
In work presented elsewhere a dis-
cussion of alternative specific models
for the late heavy bombardment is pre-
sented (Wetherill 1975b). The most
satisfactory of these are derivation of
the impacting bodies from residual
planetesimals in the vicinity of Uranus
and Neptune, possibly the proto-
cometary cloud, and derivation from
residual Mars-crossing bodies. Both of
these models are compatible with
probable concentrations and size dis-
tributions of mass in these regions of
the solar system 4 billion years ago. In
both cases Roche limit disruption pro-
vides a mechanism for producing a
simultaneous late heavy bombardment
similar to that suggested by the crater-
ing record.
The Mars-crossing source has the
property of requiring a different crater-
ing time scale for Mars than for the
moon and Mercury. A difficulty with
the Mars-crossing model is the re-
quirement of a rather special initial
distribution of Mars-crossing orbits to
obtain the sharp cut-off in lunar bom-
164
CARNEGIE INSTITUTION
bardment about 3.9 billion years ago.
Further detailed treatment of the
Uranus-Neptune source requires de-
velopment of new theoretical tech-
niques for proper handling of dynami-
cal problems in the outer solar system.
Concluding remarks. The central
question of whether it is dynamically
permissible to extend to other planets
the late heavy bombardment observed
and dated on the moon can be an-
swered. The answer is yes: Bodies in a
wide class of orbits will have very simi-
lar probabilities for impacting the
moon and terrestrial planets; there are
at least two "storage places" that are
sufficiently long-lived; the mass re-
quirements are not excessive; and
there is a dynamically plausible mech-
anism for producing simultaneous
episodes of heavy bombardment on all
the terrestrial planets.
Attention can then be shifted from
the question of whether this hypothesis
of simultaneous late heavy bombard-
ment is permissible to the much more
difficult question of whether or not it
actually occurred.
Positive evidence for the hypothesis
may some day come from radiometric
dating of impact-related rocks from
other terrestrial planets. At present
the only bodies in the solar system
other than the earth and the moon
which can be dated in this way are the
parent bodies of the meteorites. As re-
viewed elsewhere (Wetherill, 1974),
there is considerable uncertainty con-
cerning the location of these parent
bodies in the solar system. It seems
likely that at least most differentiated
meteorites-achondrites, irons, and
stony-irons are of asteroidal origin,
even though skepticism on this matter
would prove difficult to refute. The
projectiles of the late heavy bombard-
ment should have spent much of their
time in the asteroid belt and thereby
increased the impact rates on the par-
ent bodies of these differentiated
meteorites, and this should be reflected
in the meteorite age measurements.
Even if it occurred, this increased
impact rate in the asteroid belt may
not have been very pronounced because
of the high background collision rate in
this region of the solar system. For
example, there are at present about 30
asteroids equal to or larger than the
Imbrium planetesimal (R = 50 km). If
the late heavy bombardment involved
an additional 100 projectiles of this
size, the number of such bodies would
be increased by only a factor of 4. The
duration of the bombardment would be
~ 30-100 m.y., and the integral flux on
the asteroids would represent primar-
ily impacts unrelated to the late heavy
bombardment. It may also prove dif-
ficult to distinguish between impact
metamorphism and melting, and simi-
lar rocks resulting from internal heat-
ing of the meteorite parent bodies. At
present there is some evidence for dif-
ferentiated meteorite ages in the vicin-
ity of 3.8 b.y. (e.g., Burnett and Was-
serburg, 1967; Papanastassiou et al.,
1974). However, it is premature to say
whether these data argue for or against
a simultaneous late heavy bombard-
ment in the inner solar system.
The Earth-Crossing Apollo Objects
The study of meteorites is central to
geochemical research. The remarkable
similarity in the concentration of the
abundant nonvolatile elements (e.g.,
Mg, Si, Fe, Na, Al, Ca) in carbonaceous
and chondritic meteorites and the sun
shows that the chondrites represent
solar system material that has under-
gone only limited chemical processing.
Therefore, it is likely that chondritic
ratios such as K/Rb, Fe/Ni, and Mg/Fe
represent an approximation to funda-
mental solar system values for these
quantities rather than being more or
less random numbers associated with
special types of rocks. Also, the parti-
tion of trace elements between silicate,
metal, and sulfide phases in meteorites
is a principal basis for the geochemical
classification of the elements as
lithophile, chalcophile, and sidero-
phile. The primitive nature of meteor-
ites is indicated by their —10* yr I-Xe
formation intervals, and therefore the
DEPARTMENT OF TERRESTRIAL MAGNETISM
165
age of the solar system can be inferred
from the —4.6 x 10 9 yr age of most
meteorites. Consequently, it is quite
likely that isotopic inhomogeneities
such as those found in neon (Black and
Pepin, 1969), oxygen (Clayton and
Mayeda, 1975), and magnesium (Lee
and Papanastassiou, 1974) can be at-
tributed to nebular or even presolar
system phenomena.
For these and other reasons, our un-
derstanding of solar system history de-
pends heavily on meteoritic data. In-
valuable as these data have been, their
full importance cannot be realized
until we identify the actual bodies in
the solar system of which the meteor-
ites are fragments. Our understanding
of this problem has advanced consid-
erably during the past decade. To a
large extent these advances have made
more clear the intricacies of the prob-
lem and have shown that simple solu-
tions, whether or not they are "gener-
ally accepted," may be incorrect or at
least inconclusive. The present status
of the general problem has been re-
viewed (Wetherill 1974). It is con-
cluded therein that the present sources
of meteorites are probably multiple: It
is plausible that both comets and as-
teroidal bodies of several kinds con-
tribute to the flux of meteorites on the
earth.
During the past year attention has
been focused on a particular class of po-
tential meteorite source, the kilome-
ter-sized interplanetary bodies known
as Apollo objects. These are earth's
nearest neighbors in space; one will
pass within the orbit of the moon about
once in 50 years. The name of the first
of these bodies to be discovered, 1862
Apollo, has been used to give the name
"Apollo objects" to the entire class. The
common association of the term "as-
teroid" with this class may be mislead-
ing. Rather than being related to the
true asteroids, primarily confined to
orbits between Mars and Jupiter, these
Apollo "asteroids" may actually be the
exposed nonvolatile cores of comets,
remaining after their icy mantles have
been evaporated by solar radiation
(Opik, 1963).
There is no doubt that both Apollo
objects and their fragments impact the
earth. Their elliptical orbits extend
into the asteroid belt, and while in this
portion of their orbits they cannot es-
cape collisions with the asteroidal col-
lision debris present in that region
(Wetherill, 1967). The larger collision
fragments derived from the Apollo ob-
jects will typically have velocities less
than 100 m/s relative to the body of
which they are fragments (Gault et al.,
1963). They will travel in orbits very
similar to that of their Apollo object
source, and therefore the orbits of the
fragments will also cross the orbit of
the earth. Precession of the argument
of perihelion will cause orbits that
cross earth's orbit to evolve on a time
scale of ~10 4 years into orbits which
intersect earth's orbit. The earth will
be actually present at the point of in-
tersection a calculable fraction of the
time, and an impact with the earth will
result. Whether or not we have meteor-
ites in our collections which are frag-
ments of Apollo objects depends upon
whether the total annual number of
fragments impacting the earth is suffi-
ciently large to ensure their being rep-
resented among the —2000 meteorites
that have been found. In addition, it is
necessary that the fragments have suf-
ficient strength to survive passage
through the earth's atmosphere at the
entry velocities of typical meteorites
(11-20 km/s).
In contrast, fragments of most belt
asteroids cannot impact the earth.
Whereas the Apollo object fragments
when produced are already in earth-
crossing orbit, fragments of typical as-
teroids ejected at —100 m/s following
collisions will remain in the asteroid
belt. Only if they are ejected at very
high velocity ( — 5 km/s) will they be
placed directly into earth-crossing.
However, any reasonable partition of
this kinetic energy into internal de-
grees of freedom will melt or pulverize
the fragment, and it will no longer re-
166
CARNEGIE INSTITUTION
semble typical relatively unshocked
meteorites. The absence of lunar rocks
in our meteorite collections contributes
empirical evidence for this conclusion.
Special mechanisms have been pro-
posed which may permit removal of a
small quantity of collision fragments
from the asteroid belt without these
shock effects. The most promising of
these involve proximity of the asteroid
to an orbit that is resonant with the
motion of Jupiter (Williams, 1973;
Zimmerman and Wetherill, 1973).
These also constitute promising
meteorite sources, particularly for dif-
ferentiated meteorites such as irons.
In recognition of their prima facie
candidacy as meteorite sources a quan-
titative evaluation of the mass yield,
lifetime, and orbital characteristics of
meteorites derived from Apollo objects
was carried out a few years ago
(Wetherill and Williams, 1968). This
work showed that if the mass yield is
calculated from the approximately 20
observed Apollo objects alone, it is
much too low. Estimates were given of
the number of unobserved similar ob-
jects required in order for the mass
yield to be adequate to produce a major
part of the observed meteorite flux. Al-
though the fairly large number re-
quired could not be ruled out on obser-
vational grounds, an assumption that
there actually were so many would
have been excessively conjectural.
Extension of an approach first used
by Whipple (1967, 1973) now shows
that the total number of these bodies
larger than about 500 meters in radius
is about an order of magnitude higher
than most previous estimates. This ap-
proach makes use of the fact that no
Apollo object has been accidentally dis-
covered twice, even though the mode of
discovery approximates a random
search procedure. This would obviously
be an improbable result if there were
only a few more bodies than actually
observed. More quantitatively, this
fact has been inverted to estimate the
most probable number of Apollo ob-
jects. This procedure requires estab-
lishing an upper limit on the number of
Apollo objects; otherwise the most
probable number would be arbitrarily
large. Rather good upper limits can be
set in two ways. Shoemaker et al.
(1975) have used completeness of
search calculations to show that it is
unlikely there are more than 1200 of
these bodies. Impacts of these bodies
will produce lunar and terrestrial crat-
ers about 10 km in diameter. The small
number of these craters on surfaces 3
billion years in age shows that
Shoemaker's upper limit is if anything
too high. Thus bounded, the inverse
problem can be solved, with the result
that it is 95% probable there are more
than 200 of these bodies; a value com-
patible with all the evidence is about
600. This is in agreement with the re-
cent estimate of 800 ±400 given by
Shoemaker et al. based on complete-
ness of search.
Use of this higher abundance, to-
gether with more recent results on as-
teroid diameters and cratering me-
chanics leads to a revision of the esti-
mates of meteorite yield given previ-
ously. The total quantity of earth-
impacting Apollo fragmentation debris
in the meteorite size range ( 10 2 - 10 ,; g)
is now estimated to be about 2 x 10 8
g/yr. This is in good agreement with
estimates of the actual impact rate, as
determined by observations of fireballs
with the Prairie Network (McCroskey,
1968) and lunar seismometry (Duen-
nebier et al., 1975). Thus, provided
these fragments have sufficient
strength to penetrate the atmosphere,
it may be that most meteorites are de-
rived from these bodies.
If so, what kind of meteorites are
they likely to be? Good spectrophoto-
metric data have been obtained on one
Apollo object, 1685 Toro (Chapman et
al., 1973). They indicate a surface com-
position similar to the most abundant
class of meteorites: the metamorphosed
ordinary chondrites. Albedo and lower
resolution spectral data suggest this is
DEPARTMENT OF TERRESTRIAL MAGNETISM
167
the case for several other Apollo ob-
jects.
Discrepancies exist between the ob-
served distribution of meteorite
radiants and time of fall and that pre-
dicted for fragments of Apollo objects.
Possible but by no means assured ways
to remove these discrepancies have
been studied and discussed (Wetherill,
1976).
The short (~10 7 yr) dynamical
lifetime of Apollo objects requires a
source to maintain the supply through-
out solar system history. Various pos-
sible asteroidal sources have been in-
vestigated. Several mechanisms for
removing kilometer-sized asteroid
fragments from the asteroid belt are
qualitatively acceptable. However,
quantitatively these mechanisms fail
by at least an order of magnitude. An
adequate theory of asteroidal origin
requires a mechanism by which ~5% of
all the Apollo-sized fragments being
produced in the asteroid belt undergo
the necessary ~5 km/s velocity change
to cause them to become earth-
crossing. No such efficient mechanisms
are known at present. An alternative is
that most Apollo and many Amor ob-
jects are the cores of comets that have
lost their volatiles by solar heating
(Opik, 1963). Observational evidence
exists that there actually are a suffi-
cient number of such cores, and it is
believed that this is a likely possibility.
The principal uncertainty in relating
chondritic meteorites to comets is the
possibility that the Apollo material
does not have the mechanical strength
characteristic of ordinary chondrites.
If the foregoing chain of reasoning is
followed, then ordinary chondritic
meteorites are fragments of comets. In
spite of the fact that this conclusion
does not fit well with our present
thinking regarding the chemical his-
tory of the early solar system, these in-
vestigations indicate that there is a
problem to which this conclusion may
be a solution.
References
Aggarwal, H. R. and V. R. Oberbeck, Roche
limit of a solid body, Astrophys. J., 191,
577-588, 1974.
Arnold, J. R., The origin of meteorites as
small bodies, 2, the models; 3, general
considerations, Astrophys. J., 141, 1536-
1547, 1965.
Baldwin, R. B, The Measure of the Moon, U.
Chicago Press, p. 488, 1963.
Black, D. C, and R. O. Pepin, Trapped neon
in meteorites — II, Earth Planet. Sci. Lett.,
6, 395-405, 1969.
Blake, M. L., and G. J. Wasserburg, Crater-
ing and cosmogenic nuclides, Geophys.
Res. Lett., 2, A11-A19, 1976.
Burnett, D. S., and G. J. Wasserburg, Evi-
dence for the formation of an iron meteor-
ite at 3.8 x 10 9 , Earth Planet. Sci. Lett., 3,
137-147, 1967.
Cameron, A. G. W., Accumulation pro-
cesses in the primitive solar nebula,
Icarus, 18, 407-450, 1973.
Cameron, A. G. W., and W. R. Ward, The
origin of the moon. In Lunar Science, 7,
pp. 120-122, Lunar Science Institute,
Houston, 1976.
Chapman, C. R., Cratering on Mars: 1.
Cratering and obliteration history,
Icarus, 22, 272-291, 1974.
Chapman, C. R., T. B. McCord, and C. Piet-
ers, Minor planets and related objects X.
Spectrophotometry study of the composi-
tion of (1685) Toro, Astron. J., 78, 502-
505, 1973.
Clayton, R. N., and T. K. Mayeda, Genetic
relations between the moon and meteor-
ites, Proc. 6th Lunar Sci. Conf, pp.
1761-1769, Pergamon Press, New York,
1975.
Duennebier, F., J. Dorman, D. Lammlein,
G. Latham, and Y. Nakamura, Meteoroid
flux from passive seismic experiment
data, Proc. 6th Lunar Sci. Conf, pp.
2417-2426, Pergamon Press, New York,
1975.
Gault, D. E., Saturation and equilibrium
conditions for impact cratering on the
lunar surface: criteria and implications,
Radio Science, 5, 273-291, 1970.
Gault, D. E., E. M. Shoemaker, and H. J.
Moore, Spray ejected from the lunar sur-
face by meteoroid impact, NASA TN
1767, 39, 1963.
Gerstenkorn, H., The earliest past of the
earth-moon system, Icarus, 11, 189-207,
1969.
168
CARNEGIE INSTITUTION
Gurevich, L. E., and A. I. Lebedinskii, For-
mation of the planets, Izvestia Akad.
Nauk USSR, Seriya Fizich., 14, 765-799,
1950.
Hallam, M., and A. H. Marcus, Stochastic
coalescence model for terrestrial planet-
ary accretion, Icarus, 21, 66-85, 1974.
Hanks, T. O, and D. L. Anderson, The early
thermal history of the earth, Phys. Earth
Planet. Int., 2, 19-29, 1969.
Hartmann, W. K., and D. R. Davis,
Satellite-size planetesimals and lunar
origin, Icarus, 24, 504-515, 1975.
Jeffreys, H., The relation of cohesion to
Roche's limit, Mon. Notic. Roy. Astron.
Soc, 107, 260-262, 1947.
Kaula, W. M., Dynamical aspects of lunar
origin, Rev. Geophys. Space Phys., 9,
227-248, 1971.
Kaula, W. M., and P. E. Bigeleisen, Early
scattering by Jupiter and its collision ef-
fects in the terrestrial zone, Icarus, 25,
18-33, 1975.
Kaula, W. M., and A. W. Harris, Dynamics
of lunar origin and orbital evolution,
363-371, Rev. Geophys. Space Phys., 13,
1975.
Lee, T., and D. A. Papanastassiou, Mg
isotopic anomalies in the Allende meteor-
ite and correlation with O and Sr effects,
Geophys. Res. Lett., 1, 225-228, 1974.
Marcus, A. H., Positive stable laws and the
mass distribution of planetesimals,
Icarus, 4, 267-272, 1965.
Marcus, A. H., Speculations on mass loss by
meteoroid impact and formation of the
planets, Icarus, 11, 76-87, 1969.
McCrosky, R. E., The distribution of large
meteoritic bodies, Smithsonian As-
trophys. Observ. Spec. Rep., 280, 13 pp.,
1968.
Mitler, H. E., Formation of an iron-poor
moon by partial capture, or: yet another
exotic theory of lunar origin, Icarus, 24,
256-268, 1975.
Mizutani, H. T., T. Matsui, and H.
Takeuchi, Accretion process of the moon,
The Moon, 4, 476-489, 1972.
Murray, B..R. G. Strom, N. J. Trask, and D.
E. Gault, Surface history of Mercury: Im-
plications for terrestrial planets, J.
Geophys. Res., 80, 2508-2514, 1975.
Neukum, G., B. Konig, and J. F. Arkani-
Hamed, A study of lunar impact crater
size distributions, The Moon, 12, 201-
229, 1975.
O'Keefe, J. D., and T. J. Ahrens, Shock ef-
fects from a large impact on the moon,
Proc. 6th Lunar Sci. Conf., 2831-2844,
1975.
O'Keefe, J. D., and T. J. Ahrens, The high
speed ejecta from a meteorite impact and
planetary accretion (abstract), EOS, 57,
21 A, 1976.
Opik, E. J., Roche's limit; Rings of Saturn,
Irish Astron. J., 1, 25-26, 1950.
Opik, E. J., Collision probabilities with the
planets and the distribution of the in-
terplanetary matter, Proc. Roy. Irish
Acad.,54A, 165-199, 1951.
Opik, E. J., Survival of comet nuclei and
the asteroids, Advan. Astron. Astrophys.,
2, 219-262, 1963.
Opik, E. J., Sun-grazing comets and tidal
disruption, Irish Astron. J., 7, 141-161,
1966.
Opik, E. J., Comments on lunar origin,
Irish Astron. J., 10, 190-238, 1972.
Papanastassiou, D. A., R. S. Rajan, J. C.
Huneke, and G. J. Wasserburg, Rb-Sr
ages and lunar analogs in a basaltic
achondrite; implications for early solar
system chronologies, In Lunar Science V,
pp. 583-585, Lunar Sci. Institute, Hous-
ton, 1974.
Roche, E. A., Academie des Sciences et
Lettres de Montpellier, Memoires de la
Section des Sciences, 2, 243-262, 1847.
Ruskol, E. L., The origin of the moon I, for-
mation of a swarm of bodies around the
earth, Sov. Astron. AJ, 4, 657-668, 1960.
Ruskol, E. L., On the origin of the moon II,
the growth of the moon in the circumter-
restrial swarm of satellites, Sov. Astron.
AJ, 7, 221-227, 1963.
Ruskol, E. L., The origin of the moon III,
some aspects of the dynamics of the cir-
cumterrestrial swarm, Sov. Astron. AJ,
15, 646-654, 1972.
Ruskol, E. L., On the origin of the moon, In
Proceedings of the Soviet-American Con-
ference on Cosmochemistry of the Moon
and Planets, pp. 638-644, Nauka, Mos-
cow, 1975.
Ruskol, E. L., E. V. Nikolaevo, and A. S.
Syzdykov, Dynamical history of coplanar
two-satellite systems, The Moon, 12,
11-18, 1975.
Safronov, V. S., Accumulation of the
earth-group planets, Voprosy Kos-
mogonii, 7, 59-65, 1960.
Safronov, V. S., Evolution of the Proto-
planetary Cloud and Formation of the
Earth and Planets, Nauka, Moscow,
translated for NASA and NSF by Israel
DEPARTMENT OF TERRESTRIAL MAGNETISM
169
Program for Scientific Translations,
(1972); 1969.
Safronov, V. S., Time scale of the formation
of the earth and planets and its role in
their geochemical evolution, In Proceed-
ings of the Soviet- American Conference on
the Cosmochemistry of the Moon and
Planets, Moscow, June 4-8, 1974, pp.
624-629, Nauka, Moscow, 1975.
Safronov, V. S., and E. V. Zvyagina, Rela-
tive sizes of the largest bodies during the
accumulation of the planets, Icarus, 10,
109-115, 1969.
Sekiguchi, N., On the fissions of a solid
body under influence of tidal force; with
application to the problem of twin craters
on the moon, The Moon, 1, 429-439, 1970.
Shoemaker, E. M., R. J. Hackman, and R.
E. Eggleton, Interplanetary correlation of
geologic time, Adv. Astronaut. Sci., 8,
70-89, 1962.
Shoemaker, E. M., E. F. Helin, and S. L.
Gillett, Populations of the planet-crossing
asteroids. Abstract of paper presented at
the International Colloquium of Planet-
ary Geology in Rome, Italy, September
22-30, 1975.
Singer, S. F., Origin of the moon by tidal
capture and some geophysical conse-
quences, The Moon, 5, 206-209, 1972.
Smith, J. V., Origin of the moon by disin-
tegrative capture with chemical dif-
ferentiation followed by sequential accre-
tion, Lunar Science, V, 718-720, Lunar
Science Institute, Houston, 1974.
Soderblom, L. A., C. D. Condit, R. A. West,
B. M. Herman, and T. J. Kreidler, Mar-
tian planetwide crater distributions: Im-
plicationsfor geologic history and surface
processes, ^Icarus, 22, 239-263, 1974.
Tera, F., D. A. Papanastassiou, and G. J.
Wasserburg, Isotopic evidence for a ter-
minal lunar cataclysm, Earth Planet. Sci.
Lett, 22, 1-21, 1974.
Trask, N. A., and J. E. Guest, Preliminary
1 geologic-terrain map of Mercury, J.
Geophys. Res., 80, 2461-2477, 1975.
Urey, H. C, Primary and secondary objects,
J. Geophys. Res., 64, 1721-1737, 1959.
Weidenschilling, S. J., A model for accre-
tion of the terrestrial planets, Icarus, 22,
426-435, 1974.
Weidenschilling, S. J., Mass loss from the
region of Mars and the asteroid belt,
Icarus, 26, 361-366, 1975.
Wetherill, G. W., Collisions in the asteroid
belt, J. Geophys. Res., 72, 2429-2444,
1967.
Wetherill, G. W., Time of fall and origin of
stone meteorites, Science, 159, 79-82,
1968.
Wetherill, G. W., Cometary vs. asteroidal
origin of chondritic meteorites, In Physi-
cal Studies of the Minor Planets, T.
Gehrels, ed., 447-460. NASA SP-267,
1971.
Wetherill, G. W., Solar system sources of
meteorites and large meteoroids, Ann.
Rev. Earth Planet. Sci., 2, 303-331, 1974.
Wetherill, G. W., Possible slow accretion of
the moon and its thermal and petrological
consequences, pp. 184-188, Conf. on ori-
gin of mare basalts and their implications
for lunar evolution, Lunar Science Insti-
tute, Houston, Nov. 17-19, 1975, 1975a.
Wetherill, G. W., Late heavy bombardment
of the moon and terrestrial planets, Proc.
6th Lunar Sci. Conf, 1539-1559, Perga-
mon Press, 1975b.
Wetherill, G. W., The role of large bodies in
the formation of the earth and moon, In
press, Proc. 7th Lunar Sci. Conf., Perga-
mon Press, 1976.
Wetherill, G. W., and J. G. Williams,
Evaluation of the Apollo asteroids as
sources of stone meteorites, J. Geophys.
Res., 73, 635-648, 1968.
Whipple, F. L., The meteoritic environment
of the moon, Proc. Roy. Soc. London,
296A, 304-315, 1967.
Whipple, F. L., Note on the number and
origin of Apollo asteroids, The Moon, 8,
340-345, 1973.
Wilhelms, D. E., Comparison of Martian
and lunar multi-ringed circular basins, J.
Geophys. Res., 78, 4084-4095, 1973.
Williams, J. G., Secular perturbations in
the solar system, Ph.D. dissertation,
U. California, Los Angeles, 1969.
Williams, J. G., Proper elements, families,
and belt boundaries, In Physical Studies
of the Minor Planets, ed. T. Gehrels,
177-181, NASA SP-267, 1971.
Williams, J. G, Meteorites from the as-
teroid belt? EOS, 54, 233, 1973.
Zimmerman, P. D., and G. W. Wetherill,
Asteroidal source of meteorites, Science,
182, 51-53, 1973.
Zvyagina, Y. V., G. V. Pechernikova, and
V. S. Safronov, A qualitative solution of
the coagulation equation taking into ac-
count the fragmentation of bodies, As-
tron. Zh. 50, 1261-1273, 1973.
170
CARNEGIE INSTITUTION
Particle Tracks and
Micro-Impact Features in
Grains from Gas-Rich
Meteorites
G. Poupeau* and R. S. Rajan
Introduction
Gas-rich meteorites are extraterres-
trial breccias made of single crystals
and lithic fragments once exposed to
free space before compaction, as evi-
denced by the presence in individual
grains of tracks from solar flare, iron
group nuclei (Lai and Rajan, 1969; Pel-
las et al., 1969) and solar wind 4 He
(Poupeau et al., 1974). From the aniso-
tropy of solar flare irradiation effects in
single crystals, it was inferred that the
irradiation occurred in a regolith on
the surface of a parent body (Poupeau
and Berdot, 1972; Rajan, 1974). From
analogy with the lunar soil samples,
we should expect that impact features
such as microcraters and glassy
splashes will also be found on crystal
surfaces of gas-rich meteorites. This
expectation was indeed confirmed
(Brownlee and Rajan, 1973) when such
features were found on solar flare ir-
radiated glass balls of the gas-rich
howardite Kapoeta. Similar observa-
tions have recently been reported for
carbonaceous chondrites (Goswami et
al., 1976). Radiometric studies of indi-
vidual clasts from a few gas-rich
meteorites have shown that these brec-
cias could have been compacted as
early as —3.5 AE ago in the case of
Kapoeta (Papanastassiou et al., 1974;
Rajan et al. 1976), -3.7 AE for Malvern
(Rajan et al, 1975) to -4.3 AE for
Bununu (Rajan et al., 1975). Fission
track evidence also suggests that the
carbonaceous chondrite Murchison was
probably compacted —4.2 AE ago
(Kothari and MacDougall, 1975).
So far, extensive studies of the solar
flare tracks/microcraters record in
single grains of extraterrestrial re-
* McDonnell Center for the Space Sciences of
Washington University, St. Louis, Missouri
63130. On leave from Centre National de la Re-
cherche Scientifique, Paris, France.
golith have been limited to the lunar
soils (Poupeau et al., 1975; Poupeau
and Johnson, 1976) and the carbonace-
ous chondrite Murchison (Goswami et
al., 1976). We are extending these
studies to a number of gas-rich meteor-
ites, with the main purposes of ( 1) de-
fining the micrometeorite mass distri-
bution over different geological times
as inferred from microcrater meas-
urements, and (2) studying the tem-
poral variations of the micrometeorite
flux relative to the solar flare flux and
ultimately, in collaboration with E.
Zinner, to the solar wind flux. The
solar wind record can be studied in
single grains with the help of an ion
probe (Zinner et al., 1976). Finally, the
combined study of glassy splashes as
well as microcraters and tracks on
single meteoritic grains should give
some insight into the regolith
dynamics of their parent body.
In this report we present results of a
search for microcraters and other im-
pact features (glassy discs/splashes) as
well as the measurement of the
cosmic-ray track records in the gas-rich
meteorite Kapoeta.
Sampling and Experimental
Procedures
Gas-rich meteorites are known in
several groups of stony meteorites: or-
dinary and carbonaceous chondrites as
well as achondrites such as howardites
and aubrites. We have selected for our
work, on the basis of their high content
of solar wind gases and abundance of
track-rich grains, one meteorite from
some of these groups: the howardite
Kapoeta and the aubrite Khor Temiki.
For comparison, we have also included
the carbonaceous chondrite Murchison.
The samples of Kapoeta and Murchison
were single fragments and weighed
— 0.5 and —0.8 grams, respectively.
Our sample of Khor Temiki is an
aliquot of a 92-150 ,um sieved powder
previously used by Eberhardt et al.
(1965) for rare gas analysis.
In order to study the impact record of
micrometeorites on single grains, we
DEPARTMENT OF TERRESTRIAL MAGNETISM
171
have disaggregated the fragments of
Kapoeta and Murchison into their orig-
inal grains. It is very important that
during this process the "external" sur-
faces of the crystals and chondrules are
preserved. Although gentle crushing
may partly preserve the external sur-
face of crystals, we have increased the
recovery of original surfaces by ther-
mally induced disaggregation.
For this purpose at DTM we have de-
signed and built an automated device
for gentle disaggregation of the
meteorite samples. The sample is first
evacuated to ~1 fx and filled with
deionized water. The tube containing
the sample is then alternately cycled
between liquid nitrogen and water at
45°C. A schematic diagram of the de-
vice is shown in Fig. 27. The method is
very effective in recovering grains with
their original surfaces intact for grains
greater than 50 /x, as deduced from
scanning electron microscope (SEM)
observations.
The Khor Temiki powder was pre-
pared by Eberhardt et al. (1965) from
gentle crushing of a fragment of this
meteorite. The low degree of compac-
tion of this meteorite allows an easy
separation of its constituent crystals. It
has been shown from the presence of
both solar wind 4 He and solar flare
VH-ion tracks in single crystals that
original surfaces have in fact survived
the crushing process at least partially
(Poupeau^ a/., 1974).
Impact features on the surface of ex-
traterrestrial grains consist of micro-
craters and a variety of glassy splashes
(cf. Blanford et al., 1974). The presence
of microcraters, in either meteorites or
lunar material, seems to be limited to
grains irradiated by the solar flares
(Brownlee and Rajan, 1973; Goswami
et al., 1976; Poupeau and Johnson,
1976). On the contrary, it has been
shown that in lunar soils, glassy
splashes were present on almost all
crystals (Poupeau and Johnson, 1976).
We therefore applied different proce-
dures for the search of microcraters
and glassy splashes on meteoritic crys-
tals. For glassy splashes, a series of
crystals chosen randomly among those
exhibiting euhedral to nearly euhedral
shapes were observed in an SEM. For
microcraters, as the abundance of solar
flare irradiated grains is usually
rather low, a selection had to be made
on the basis of nuclear track data.
Therefore, series of crystals were
mounted in epoxy, polished, and
etched. Only crystals having solar flare
TIMER
0- 300 SEC
LEFT LIMIT
+ REV. SWITCH
LIQUID N SUPPLY
1 10 VAC
1 10 VAC
REVERSIBLE
MOTOR
DEIONIZED WATER
RIGHT LIMIT
+ REV. SWITCH
IMMERSION HEATER
: -H«
45°C WATER
Fig. 27. An automated device for gentle disaggregation of meteorites (and lunar samples).
172
CARNEGIE INSTITUTION
tracks were extracted from the epoxy
and their remaining external surface
studied for microcraters in an SEM.
Results and Discussions
We have studied for impact features
— 70 crystals and glassy objects from
the three meteorites Kapoeta, Khor
Temiki, and Murchison. The search for
microcraters and glassy splashes was
carried out under a standard 5000 x
magnification scan (Poupeau et al.,
1975) in order to make a meaningful
comparison with the results from the
lunar samples. The results are reported
in Table 5, where they are compared
with other available lunar and
meteoritic data. All our measurements
(including the data on lunar soils) have
been obtained on crystals in the
— 100-150 /xm range.
Of the 45 crystals studied from
Kapoeta, Khor Temiki, and Murchison,
none had any detectable microcraters.
Brownlee and Rajan (1973) had found
microcraters in 2 out of the 9 glassy
spherules from Kapoeta. We scanned
16 more spherules, but no additional
microcraters (^0.2 /xm) were detected.
Lunar and meteoritic microcraters
present the same general morphology
illustrated in Fig. 28. Figure 28(a and
b) shows cratered glass balls from the
lunar breccias 15086 and Kapoeta, re-
spectively. The largest microcraters
found in the meteoritic spherule (Fig.
28b) have diameters of 9 and 6 fxra, re-
spectively Brownlee and Rajan, 1973).
One of the above craters is shown in
Fig. 28(d), and a crater found in a lunar
glass sphere from the regolith breccia
15086 is shown for comparison in Fig.
28(c). Both craters on Fig. 28(c and d)
present characteristic spall zones sur-
rounding the central pits. Submicron
craters in both lunar and meteoritic
grains are typically characterized by
raised rims around the pits (Fig. 28, e
and f). By analogy to the lunar soil
studies, we have classified the thin
discs of 5£2 jam diameter (the "pan-
cakes" of Blanforde^ al., 1974) from the
thicker glassy splashes on the surface
of crystals (Fig. 28f). No glassy discs
have yet been found on any meteoritic
crystal, but some are possibly present
on the surface of glassy spherules.
Thicker glassy splashes are defi-
nitely present on glassy spherules in
TABLE 5. Cosmic-Ray Tracks, Microcraters, and Glassy Discs on Crystals from
Lunar Soils and Gas-Rich Meteorites
Fraction of Crystals with:
Track
Glassy
Mineral
Densities
Microcraters
Discs
Sample
Phase
3= 10*/cm 2
(N)t
ssO.Lum
s*2/um dia.
(AD*
Lunar Soils'"
12033
Ft
0.03
(156)
0.07
(14)
14141
F
0.35
( 60)
0.14
0.74
(19)
Mature Soils
F
2*0.8
O200)
0.35-0.67
&0.90
O200)
Meteorites
Kapoeta
F.Pyt
0.13
( 65)
(15)
"glass" balls
0.20
( 15)
0.08
-0.04
(25)
Khor Temiki
Py
0.12 ,2)
(1203)
(15)
Murchison
Olt
(15)
01
0.01-0.05 I: "
(3000)
-0.001
N.D.
(20)
*(/V) = total number of observed crystals.
tF = Feldspar; Py = pyroxene; 01 = olivine; N.D. = not determined.
(1) Poupeau and Johnson, 1976.
(2) Poupeau et al., 1974.
(3) Goswami et al., 1976.
DEPARTMENT OF TERRESTRIAL MAGNETISM
Lunar Meteoritic
173
21L
Fig. 28. Microcraters and glassy splashes on meteoritic and lunar glassy spherules and crystals,
as observed under a scanning electron microscope (SEM). (a),(c) Glassy spherule from the Apollo 15
regolith breccia 15086. (b),(d) Glassy spherules from Kapoeta meteorite, (e) A feldspar crystal from
the lunar soils. Flat glassy discs ("pancakes") are indicated by arrows, (f) A glassy splash on the
surface of a spherule from Kapoeta. A submicron crater is indicated by the arrow.
Fig. 28(e). On crystal surfaces, we have
not yet found any features that can be
unambiguously attributed to glassy
splashes. This may be in part due to the
large background of adhering grains on
the crystal surface, many of which are
in fact embedded.
Our track density (= P) meas-
urements on 40 feldspar grains from a
single centimeter-sized fragment give
well-defined peak in the range of 1.5 ±
0.8) x 10'Vcm 2 , containing 29 crystals.
The above peak is well understood as
due to the stopping iron nuclei from the
galactic cosmic rays during the
cosmic-ray exposure age of the meteor-
ite. An example of such a crystal is
shown in Fig. 29(c). Six crystals had p
values in the range of 2.4-4.0 x 10 ( V
cm 2 , which is possibly due to a some-
174
CARNEGIE INSTITUTION
OPTICAL
MICROSCOPE
ELECTRON
MICROSCOPE
(b)
OPTICAL
MICROSCOPE
(c)
v..v-
* «
" * v f *
2
F
u
IO b
to
«
Y
O
b
<
CL
4
h-
\1
O
q:
2
Ld
ID
107
>
I0 6
■
«8
o°fl
o°o°
V'"---
I — I
Y-\
« \
i 1 1 1 r^— i — i — i — i — i — i — r^n 1 1 1 r
HIGH DENSITY EDGE
o TRAVERSAL I IN
KAPG-4
KAPG-4
o _
o -
88
Gal C.R = |-2XICrcm
ooooo
6. m 2
<^
-iy
j i L
J UJ-
20 60 100 \ 200 4QQ 600 / 700 740 780
Notice change in scale
DEPTH IN MICRONS
(d)
Fig. 29. Cosmic-ray tracks in Kapoeta feldspars. (a),(b) "Track-rich" crystals with track density
gradients from the edges, as viewed under an optical (a) or in a transmission electron microscope (b),
using a C-Pt replica technique (Rajan, 1974). The scale bars are 10 /j. and 2 /x and track density values
at the edges are -2 x 10>7cm 2 and ~2 x 10 ! 7cm 2 . (c) Crystal with a low homogeneous track density,
accumulated from galactic cosmic rays during the cosmic-ray exposure age of the meteorite. The track
density is ~2 X 10'Vcm 2 .
what closer location to the preatmo-
spheric surface. The remaining five
crystals have significantly higher
track densities, indicative of a precom-
paction irradiation history. While one
crystal had p = 1.5 x 10 7 /cm 2 , the re-
DEPARTMENT OF TERRESTRIAL MAGNETISM
175
maining four crystals were "track-
rich", i.e., they had track density at the
edges of 2= 10 8 /cm 2 and had character-
istic steep gradients toward the center
of the crystal, with p values at the cen-
ter of the crystal being ~2 x 10 7 /cm 2 .
An example of a track-rich grain is
shown in Fig. 29 (a and b), and the
measured track-density profile across
the crystal is shown in Fig. 29(d), indi-
cating the dramatic gradient.
Finally, we removed the track-rich
crystals from the epoxy mounts and ob-
served their remaining external sur-
faces under an SEM. No microcraters
larger than 0.15 fim (our detection lim-
it) were found in any of these four crys-
tals.
It is known from lunar soil studies
that for a grain to be cratered, being
track-rich is a necessary but not suffi-
cient condition (Poupeau and Johnson,
1976). The probability of a track-rich
grain being cratered increases with the
maturity of the soil, being essentially
zero for immature soils, evolving in a
complex way, and approaching unity
for very mature soils. Interestingly,
from the present work we find that the
meteorites lie on the extrapolation of
the same trend line exhibited by lunar
soils. Such a correlation would be sig-
nificant when confirmed with more ob-
servations on meteorites and would
suggest a common dynamical mecha-
nism operative on all the regoliths.
However, we do not yet understand the
scarcity of glassy splashes on meteori-
tic grains, since they are quite abun-
dant even on rather immature lunar
soils (see Table 5).
References
Blanford, G. E., R. M. Fruland, D. S.
McKay, and D. A. Morrison, Lunar sur-
face phenomena: solar flare track gra-
dients, microcraters and accretionary
particles, Proc. 5th Lunar Sci. Conf., 3,
2501, 1974.
Brownlee, D. E., and R. S. Rajan, Discovery
of micrometeorite craters on chondrule-
like objects from Kapoeta, Science, 182,
1341, 1973.
Eberhardt, P., J. Geiss, and N. Grogler,
Ueber die Verteilung der Uredelgase im
meteoriten Khor Temiki, Tschermaks
Mineral. Petrogr. Mitt., 10, 535, 1965.
Goswami, J. N., I. D. Hutcheon, and J. D.
MacDougall, Microcraters and solar
flare tracks in crystals from carbonace-
ous chondrites and lunar breccias, Proc.
7th Lunar Sci. Conf., in press, 1976.
Kothari, B. K., and J. D. MacDougall,
Meteoritics, 10, 428, 1975.
Lai, D., and R. S. Rajan, Observations on
space irradiation of individual crystals
of gas-rich meteorites, Nature, 223, 269,
1969.
Papanastassiou, D.A., R. S. Rajan, J. C.
Humeke, and G. J. Wasserburg, Rb-Sr
ages and lunar analogs in a basaltic
achondrite: implications for early solar
system chronologies, Lunar Science V
(C. Watkins, ed.) Lunar Science Insti-
tute, Houston, 583, 1974.
Pellas, P., G. Poupeau, J. C. Lorin, H.
Reeves, and J. Audouze, Primitive low-
energy particle irradiation of meteorite
crystals, Nature, 223, 272, 1969.
Poupeau, G., and J. L. Berdot, Irradiation
ancienne et recente des aub rites, Earth
Planet. Sci. Lett., 14, 381, 1972.
Poupeau, G., and J. Johnson, Micrometeor-
ite and solar flare maturation of lunar
soils, Earth Planet. Sci. Lett., in press,
1976.
Poupeau, G., T. Kirsten, F. Steinbrunn, and
D. Storzer, The records of solar wind and
solar flare in aubrites, Earth Planet. Sci.
Lett., 24, 229, 1974.
Poupeau, G., R. M. Walker, E. Zinner, and
D. A. Morrison, Surface exposure history
of individual crystals in the lunar re-
golith, Proc. 6th Lunar Sci. Conf, 3,
3433, 1975.
Rajan, R. S., On the irradiation history and
origin of gas-rich meteorites, Geochim,
Cosmochim. Acta, 38, 111, 1974.
Rajan, R. S., J. C. Huneke, S. P. Smith', and
G. J. Wasserburg, 40 Ar- 39 Ar chronology
of isolated phases from Bununu and
Malvern howardites, Earth Planet. Sci.
Lett., 27, 181, 1975.
Rajan, R. S., J. C. Huneke, S. P. Smith, and
G. J. Wasserburg, 40 Ar- 39 Ar chronology
of individual clasts from the Kapoeta
howardite and the origin of howardites,
submitted to Earth Planet Sci. Lett.
Zinner, E., R. M. Walker, J. Chaumont, and
J. C. Dran, Ion probe analyses of artifi-
cially implanted ions in terrestrial sam-
ples and surface-enhanced ions in lunar
sample 76215,77, Proc. 7th Lunar Sci.
Conf, 1976.
176
CARNEGIE INSTITUTION
Cosmic-ray Research
Scott E. Forbush
Cosmic-ray diurnal anisotropy and
the reversals of the sun's poloidal mag-
netic field. Variations with a "period"
of two solar cycles in the cosmic-ray
diurnal anisotropy component in the
asymptotic direction 128° east of the
sun which are related to the reversals
of the sun's poloidal magnetic field
have been discussed in previous annual
reports. The component of the diurnal
anisotropy in the asymptotic direction
90° east of the sun varies with
geomagnetic activity and exhibits a
solar cycle variation. Together these
two components account for essentially
all the variations of the annual means
of the diurnal anisotropy from 1937 to
1974, as was reported by Forbush at
the 14th International Cosmic-Ray
Conference in Munich, August 1975.
These results could only have been de-
rived from data covering the long
interval since 1937; such data are
available from three stations with
Carnegie Institution of Washington
ionization chambers. Nevertheless,
this interval is less than that covered
by two "cycles" in the variation of the
sun's poloidal magnetic field. Thus, it
is desirable to extend the observational
series for at least another two or three
years (circa 1978), which fortunately is
being achieved by the unselfish, whole
hearted, gratuitous cooperation of in-
dividuals at three stations,
Observations and reduction of data.
Cosmic-ray ionization chambers were
operated throughout the report year at
Huancayo, Peru; at Fredericksburg,
Virginia, U.S.A.; and at Christchurch,
New Zealand. It is anticipated that by
the end of 1976 the scaling and reduc-
tion of records will again be on a cur-
rent basis.
Cooperation in operation of cosmic-
ray meters. Appreciation is expressed
to the U.S. Geological Survey and the
U.S. Department of the Interior and its
staff at the Fredericksburg Geomagne-
tic Center for efficient operation of the
cosmic-ray meter there during the past
report year, and to the Government of
Peru and the Director and staff of the
Instituto GeofTsico del Peru for making
cosmic-ray records from Huancayo
available. Appreciation is also ex-
pressed to the Director and staff of the
Geophysical Observatory at Christ-
church, New Zealand, for the excellent
maintenance of the equipment there
and for the fine records obtained.
Rb-Sr Mantle Isochrons
C. Brooks, D. E. James, S.R. Hart,
and A. W. Hofmann
INTRODUCTION
The use of isotopic abundances of
strontium in volcanic rocks as tracers
of the source composition has long been
an important geochemical tool for the
study of the earth's mantle. Because of
their uniquely low 87 Sr/ Hti Sr ratios, it
has been shown, for example, that the
mid-ocean ridge volcanics must tap a
mantle source that is different from
virtually all other volcanic rocks in-
cluding those found on oceanic islands.
A difficulty encountered in the in-
terpretation of the isotopic data is that
these other groups of volcanic rocks,
though nearly always higher in K7 Sr/
K,; Sr than mid-ocean ridge basalts, tend
to have much more variable isotope
ratios, thus obscuring the isotopic dis-
tinction even between crustal and
mantle derived material.
In the two sections of this report we
attempt to obtain additional informa-
tion by using both the 8T Sr/ H,i Sr ratio
and the Rb/Sr ratio of these rocks in
conventional isochron diagrams. Be-
cause isolated occurrences of apparent
Rb-Sr isochrons with "ages" greatly in
excess of the age of the volcanic erup-
tions have been reported before, we
undertook a systematic search of the
literature and our own data for such
DEPARTMENT OF TERRESTRIAL MAGNETISM
177
pseudoisochrons. We found that pseu-
doisochrons are indeed very common in
individual groups of volcanic rocks.
Moreover there is a well-defined single
pseudoisochron for a world-wide sam-
pling of oceanic tholeiitic basalts.
We believe that these pseudoiso-
chrons are inherited from the mantle
and that the volcanic rocks therefore
contain much more specific informa-
tion about the local and world-wide
evolution of the mantle than had been
realized. For example, the world-wide
Rb-Sr isochron age of 1.6 ±0.2 b.y. for
tholeiites from oceanic islands and
mid-ocean ridges may date a major
event in the evolution of the earth's
mantle. Similarly, the more variable
and mostly younger ages derived from
isochron correlations for more localized
groups of continental volcanic rocks
are interpreted as inherited from old
subcontinental lithosphere.
To account for this heterogeneous
structure of the mantle we propose a
model for the mantle whereby diapirs
or "blobs" rise from the lower mantle
(mesosphere) through a partially mol-
ten upper mantle (asthenosphere) until
they encounter the completely solid
uppermost mantle (lithosphere). In
oceanic regions the lithosphere is rela-
tively thin (100 km or less), and the
blobs rise directly to the surface to form
oceanic islands. The world-wide mantle
isochron for these rocks is generated by
mixing of mesosphere and asthenos-
phere in different proportions, and the
age of 1.6 b.y. may date the chemical
fractionation and segregation of these
largest-scale heterogeneities of the
mantle. In continental regions the
lithosphere is thicker (up to 400 km),
and the mesospheric material does not
necessarily penetrate directly to the
surface. Rather, much of it remains
near the base of the lithosphere and is
gradually incorporated by the slowly
thickening lithosphere which grows by
freezing out of partially molten as-
thenosphere. Continental volcanism is
caused by reactivation of this hetero-
geneously accreted lithosphere, and
the mantle isochrons contain a record,
possibly distorted, of the accretion his-
tory of this lithosphere.
PART I: CONTINENTAL REGIONS
In this report we discuss apparent
Rb-Sr isochron relationships of vol-
canic (and a few intrusive igneous)
rocks, believed to be derived from the
subcontinental mantle. Most of the
rocks included in this discussion are
basalts and andesites, but others are
more exotic rock varieties, mostly of
strongly alkalic affinity (phonolites,
nephelinites, syenites, ijolites, and
shoshonites). The intrusive rocks in-
clude two suites of granites from
California. In most cases the age of the
continental crust is greater than 10 9
years, so it is reasonable to expect that
at least some of the underlying litho-
sphere is also quite old. In contrast
with the results reported in Part II (for
oceanic volcanic rocks), we have not
averaged any of the data by geographi-
cal (or any other) groupings because we
do not attempt to show (nor would we
expect to find) any meaningful world-
wide correlations, except perhaps that
from a statistical point of view rocks
with high Rb/Sr ratios would tend to
have high * 7 Sr/ Ki; Sr ratios.
The terminology adopted is as fol-
lows: Lithosphere is the rigid outer
shell of the earth which includes the
.crust and upper mantle; asthenosphere
is mobile (probably incipiently melted)
mantle material underlying the litho-
sphere; mesosphere is the rigid mantle
material underlying the asthenosphere
and constituting the bulk of the man-
tle. Subcontinental mantle is a general
term used primarily to mean litho-
spheric mantle but not necessarily ex-
cluding asthenosphere and meso-
sphere. Undepleted refers to a mantle
system with Rb/Sr and Sr-isotopic
ratios compatible with closed system
evolution for 4.5 b.y. Enriched and de-
pleted refer to systems in which present
Rb/Sr ratios are higher and lower, re-
spectively, than those required for
closed system evolution. Thus, the typ-
178
CARNEGIE INSTITUTION
ical mid-ocean ridge tholeiite comes
from partially depleted mantle and is
therefore much lower in K, Rb, Sr, Ba,
and Rb/Sr ratio than tholeiites of
otherwise similar composition from
oceanic islands and from continental
regions. Homogeneous implies that
chemical composition is uniform from
place to place, provided that the sample
is large enough to average out mineral
scale compositional variation. Hetero-
geneous is used to refer to composi-
tional variations on any scale. A man-
tle rr age" refers to the time of some
geological event in the mantle, such as
the development of large-scale
heterogeneities.
Pseudoisochrons. Because mantle-
derived magmas in continental areas
have traversed some 30-40 kilometers
of ancient radiogenic sialic crust, the
high and variable Sr-isotope ratios ob-
served in these magmas are commonly
ascribed to crustal contamination (e.g.,
Bell and Powell, 1970; Compston et al,
1968; Compston, 1974; Doe etal, 1969;
Ewart and Stipp, 1968; Faure et al.,
1972; Green and Ringwood, 1967; Hal-
pern et al., 1974; Hedge, 1966; Hurley
etal., 1966; Kudo et al., 1971; Laughlin
etal, 1972; Manton, 1968; Montignye^
al, 1969; Pushkar, 1968; Pankhurst,
1969; Scott et al, 1971). It is also possi-
ble, however, that these higher Sr
ratios are primary, inherited from sub-
continental mantle possessing higher
and more variable abundances of
radiogenic Sr. The observed range in
values would reflect lateral and/or ver-
tical heterogeneities within this man-
tle. The recent literature increasingly
favors the latter possibility, mostly due
to studies of alkalic rocks (e.g., Powell
and Bell, 1974), and volcanics of west-
ern U.S.A. where subcontinental litho-
sphere and asthenosphere are appar-
ently playing a major role in surface-
reaching magmatism (Hedge and No-
ble, 1971; Leeman, 1974, 1975; Leeman
and Manton, 1971; Mark et al, 1975;
Peterman et al, 1970). The possibility
that the Sr-isotopic ratios are inherited
directly from subcontinental mantle is
further strengthened by the presence of
ubiquitous pseudoisochrons. (We note
here that we use the term pseudoiso-
chrons or mantle isochron for any posi-
tively correlated array of data, as long
as this correlation is statistically sig-
nificant. In the conventional usage of
the term isochron, one usually requires
the data to define a straight line within
the limits of analytical error. The im-
plication of our more relaxed usage is
that the initial K7 Sr/ K,i Sr need not be
perfectly correlated for all samples).
Prior to discussing the data upon
which this paper is based, it is essential
to distinguish between the simple iso-
chron plot and the pseudoisochron plot
utilized throughout our data presenta-
tion. The isochron diagram for a group
of igneous rocks plots their individual
K7 Sr/ 8,i Sr composition as measured
today (termed present-day ratios) ver-
sus their respective Rb/Sr ratios. The
pseudoisochron diagram, however,
plots the initial 87 Sr/ 8,; Sr ratios of the
samples versus their respective Rb/Sr
ratios. In other words, on a pseudoiso-
chron plot the K7 Sr/ K(i Sr of the samples
have been individually "corrected
back" to their time of crystallization,
using independent information for the
age of crystallization. Ideally, a group
of volcanics 100 m.y. old would define a
line on an isochron diagram whose
slope corresponds to an age of 100 m.y.
The same volcanics on a pseudoiso-
chron plot, however, should define a
horizontal line (i.e., zero age) whose
intercept is the unique 87 Sr/ H(i Sr compo-
sition of the parent magma. When
these volcanics do not plot as a horizon-
tal array on the pseudoisochron plot,
then there is a hint that some precrys-
tallization history in H7 Sr/ K,i Sr has been
preserved in the rocks. If the data ac-
tually define a good linear array, the
hint is strengthened, and there is a
strong possibility that the information
given in the pseudoisochron plot per-
tains directly to the source region of
the volcanics. For many volcanic rock
associations of continental regimes this
source region is the mantle. Therefore,
DEPARTMENT OF TERRESTRIAL MAGNETISM
179
we searched the literature for pseudo-
isochrons with the aim of evaluating
the possible preservation of mantle
isotopic properties in surface-reaching
igneous rocks.
The results of the literature survey
are given in Table 6 and Table 7. Here,
each measured 87 Sr/ 8(i Sr ratio has been
corrected for the true age of the rock,
and the whole-rock initial 87 Sr/ 8li Sr
ratios then plotted vs. Rb/Sr on a
pseudoisochron diagram. Correlation
theory and regression analysis indicate
that most of these pseudoisochron have
slopes significantly different from zero
at confidence levels up to 95% (in some
cases up to 99.9%) and that they define
excess "ages" ranging from 70 m.y. to
more than 3000 m.y. Some examples
are shown in Fig. 30.
For the most part the pseudoiso-
chrons come from young volcanic ter-
ranes in which no age correction of the
measured present-day Sr isotopic com-
position is necessary. Noteworthy are
the numerous examples from western
U.S.A., where pseudoisochrons are ob-
served in rock associations of widely
differing chemical composition. Fur-
thermore, pseudoisochrons are not re-
stricted to young volcanics. Examples
are also encountered in plutonic and
hypabyssal rocks of Mesozoic age (the
California batholith, the Jurassic dol-
erites of Antarctica), Caledonian basic
igneous rocks of Scotland (Haddo
House mass), and in ancient, layered,
mafic intrusions (Usushwana Complex
of South Africa).
There are a variety of mechanisms
that could produce the pseudoisochrons
given in Table 6. These mechanisms
can be divided into crustal and mantle
types. The crustal-related processes
simply involve mixing of mantle-
derived magmas with crustal material
of higher 87 Sr/ 8,i Sr (crustal contamina-
tion). The criteria for evaluating this
process as a potential means of raising
the 87 Sr/ 8,i Sr of mantle-derived magmas
are well illustrated in studies of young
island arc volcanics (Ewart and Stipp,
1968; Pushkar, 1968). In these studies,
a hyperbolic relationship between
87 Sr/ 8,i Sr and Sr concentrations is taken
to be indicative of contamination.
Clearly, however, many forms of dis-
equilibrium melting or mixing of het-
erogeneous mantle may produce simi-
lar curves and otherwise be indistin-
guishable from crustal contamination.
Hence such relationships cannot be
considered diagnostic.
On the other hand, there is often in-
dependent evidence against crustal
contamination of mafic volcanics. We
include in this category such pertinent
features as:
1. Limited range of Sr isotopic com-
position of the volcanic rocks despite
the very large range in isotopic compo-
sition of the supposed contaminating
materials.
2. Limited range of Sr isotopic com-
position in many volcanic associations
despite a wide range in Sr concentra-
tions.
3. Persistence of normal rock com-
positions (e.g., tholeiites) despite sig-
nificant variability in isotopic composi-
tion and hence presumed wide varia-
tion in the nature and quantity of con-
taminant required.
4. Constancy of composition of dif-
ferentiates, irrespective of the nature
and amount of the supposed contamin-
ant.
5. Lack of correlation between
thickness or age of continental sialic
crust and the 87 Sr/ 86 Sr ratios observed
at the surface.
6. High "initial" 87 Sr/ 86 Sr ratios
that ought not to be influenced by crus-
tal contamination, assuming linear
mixing relations.
7. Lack of geographic correlation of
isotopic compositions for many tectonic
environments.
8. Lack of correlation between Sr
and Pb isotopic ratios.
9. 206 Pb/ 204 Pb ratios in basalts and
rhyolites which commonly show the
same range of values.
10. Paucity of included material
such as crustal xenoliths.
180
CARNEGIE INSTITUTION
'3
ss
13
*
c
«
«
C <^!>h
a 3f£
OS
CD CO_
OOCOCOCOC^tNTfCD-^OJTtiCo'co'oOOi— lOOOr- 1 (NiO Oi Oi
iomimho: H
C^co^cO^C0COCO(N^^^^C^T^^^■*c0Tt l '-l(^J^^^^cOTJ^T-^(^l'-^CC|T^
TtcDioon-^iNaiSoNiScoooSHHcoH^oo^iHOHaiooP^ai
oooot^a^05ooa5QOwt>a2^t>a5ooTt , a5ooaiioiOTt i Tt , ooi>Tj<ioinixia5
ooooooooooooooooooooooooodoooo
OO^CO^CDTOOOOOOCTicOOOOC^C^^COTtCOrJ<<N^)TtiTti(NcO < ^IC<l T *'- 1
CO rH rH rH i-H rH
+1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1
i>minaiHooa)'!i'Tfc-ioai'#iMHHioiMoam(N(Diotc3;® | M05Ti<
ooc^c^Tt H '^c<i(X>io^^Louo^cocDco^cococ^cocD»oinTt l 23cocotDco
OiOOOOOOOO'-HOOOOOOOOOOOOOOOPoOOO
OOOOOOOOOOOOOOOOOOOOOOOOCJOooOO
OOOOOOOOOOOOOOOOOOl>OOOOCT>OOOOOiO
■^0)'<tHOO^ , HQ00505t£iOOa)iOt^'<tiOOCl'*l>OOHOOIM(MOilOiOTl <
mcNCsicsicsioaoq'-icoco m >h
+1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1 +1
oooooooomooooooooot-ooooinoooooo
■^ootDNto^mHooiNOJt-b^ooHai^oaiMnHHHOooMt^
KlC0HHO0)0)00Q0NtD10'*'<tTJ<C0niNIM(NHHHHHHH
CO rH rH rH rH
•I I
'Z o>
.0} ^C Q
en OJ-S'S
."ti ^ o «
o> co ^5 -u
s $ * « $
< .| m m es
Cj . „ . » CO CO
•5 c c co o>
3 § °£^
!&&??
O CO CO g> 5
odd
PQ
CO O 113 co o to
~£
PQ
-9 :> i> J2 «P
23 bn
s
co a
rrt
cO s^
03
«
■ - d
03
^ -s
3
rt J5
ry
it
03 h
to
FT *>
Ph >
° r£
T3 Ph
03
03 a)
hn
8 3
03
C
^ ; — | a)
X '3 5
»<►? °
o x is
ijj
<£ d*u
.. Oi Q
co xl
»ftto"
rK o> o>
o>
03 O
< CO
v$
CO
03
o3 S
T) ^
X
err w
d So
so..
CO -~ — H
3^3 Ph
fe
O)
^j W\ '.£ 'u
CO
jH < g
35 cstf
o co
fi ^
111
Q 03
~|3
.3 o »
2 a g
>^ o
03 33?*
a^ O
In 03 03
< £pQ
03 3}
CO -P
rt '3
pq a
• - u
B-O
5 «
03 co
2 03
pa
Ph
bx) co
S3 PQ
Ph -w
5n °j
03.2
O 03
£«0
pq^;
CO M ^
_o "§ M
o o> co
CO i.g
r 5 " 1 O -g
t -1 ''3
.„ in 2
(D CD «
fi SO
> a. a
Ph 03 O
a J 8
13 J£'S
^ fl •-
o 9 £
ra Q «Jh
ca o> o3
^ JO
0) CO
+j a)
■3 .-a
* a
Q Q
«? co" .
CD CO
-P CD
s'S.-a
OJ CD CO
^1
CO jj"
^Ph *
^"cd
CD -
„ co
co JS
o> rs
S c
c a
oi o>
co a.
CO . .. CD
,2 'O, co'
So g
■°.a o
Ol HH ■-
D,^ oi
cq M co
hC-2^
+3 cj l— '
s-c 03 jj
cT co co'
;-, O) 0)
<<< T3T3
^ 03 03
-P u o
co co co
03 03 o3
WOO
o S d
n a fi
|
0'3
.a so
1 §S
^§13
< test
« co<
PQ PhH
•^ ^O cd
sa
< '3
?s
P -; ,^ ,^ ,x .^ .-; ,^ fe .^ -~ a
Mffii»o2!ncoo< coco gaco E go^cococococococo gcoco g,co g,
m
T)
t-H
a)
a
cd
+J
a
CD
a
o
c
CD
>
o
CO
M
C
CD
C
CO
t*H
X!
r!
S T3 .5
■oSfc
CO H
+J HH m
ChT3
a & 42
S
bc^
CO
+3 CD
3 IS
CD
HH 1
4H
C
a o)
CO
03 rC
o
• CD
co 13
II
« s
.2 &D
CO CD
co co
c
o?^
O.S
Ec CD
o
_C0
S CD
o
O -P
C CO
CO *h
CJ —■
T3 ®
c
*S -2
CD "
T3 C
o
CO
CO
CD
3 ."S
be a
a <D
CN .2
» 13
05 K
M CO
>^ ^ CO
>H CD HH
=> HO ^
CO 3
a be >
>,
co ^r
<B CD
CO
u
*" CD
g 2
09
^^
CO
cc o
CO
nr^
a
CO
.S'S
.2 -c
CO
CD CO
CD CO
0) CO
CO &
3 B
C
„ co CD
cd -e
■ t_l +J CO
2 8*3 -a
H) CO -H ^
O Cj ..-P
PQ g^ 2
I ! lis
g co C 53
> o ob 0)
> -rt CD O
c2ii a) o
c > 5* co
• J T3 3 g ■
>5ou
■ S §
T3 co -Q a,
^ CD C JH
co -a 3 co
t3 pfr ,3 s°
CPZ Ph
CO - « «
.3 (3
> <D
CD H
cd H
DEPARTMENT OF TERRESTRIAL MAGNETISM
181
TABLE 7. Reference Key to Table 6
Key No.
Reference
1
Bell and Powell (1970)
2
Church and Tilton (1973)
3
Compston et al. (1968)
4
Davies eial. (1970)
5
Ewart and Stipp (1968)
6
Faure etal. (1972)
7
Halpern etal. (1974)
8
Hedge et al. (1970)
9
Heier etal. (1970)
10
Jameses al. (1976)
11
Kistler and Peterman ( 1973)
12
Leeman (1970)
13
Leeman(1974)
14
Leeman and Manton ( 1971)
15
Nobel etal. (1973)
16
Pankhurst(1969)
17
Peterman et al. (1970a)
18
Peterman et al. (19706)
19
Powell and Bell (1970)
20
Stipp (1968)
In a detailed study on the cause of
high 87 Sr/ 8,i Sr ratios in Cenozoic ande-
sites and da cites of southern Peru,
criteria of this nature led us to reject
crustal contamination as a plausible
mechanism for either the high 87 Sr/ 8,i Sr
ratios or the pseudoisochrons observed
there (James et al., 1976). This conclu-
sion is especially remarkable when it is
noted that the Andean crust through
which these magmas have passed is
about 70 km thick, thicker than any
other known crustal section in the
world save that of the Himalayas.
Based on this synthesis, we advocate
that the hitherto durable and facile
crustal contamination explanation for
high 87 Sr/ 8i; Sr in continental volcanic
rocks be considered improbable, and
that other possible mechanisms lead-
ing to the observed isotopic composi-
tions be evaluated. We believe that
these mechanisms are all mantle re-
lated, implying that both the high Sr
ratios and the pseudoisochrons are the
result of mantle processes. This view-
point is supported by the observation
that some pseudoisochrons and high
87 Sr/ 8(i Sr ratios are observed in oceanic
regions where little or no possibility of
sialic contamination exists (Brooks et
al, 1976).
Mantle- related processes that could
lead to Rb-Sr pseudoisochrons include:
(1) mixing of heterogeneous mantle
material; (2) selective melting of heter-
ogeneous mantle material of the same
or different age; and (3) disequilibrium
melting of homogeneous mantle in
which individual mineral phases are in
isotopic disequilibrium.
Distinguishing between these pos-
sibilities is difficult, as none of the pro-
posed mechanisms leads to a uniquely
identifiable compositional property in
the rocks, other than that of Pb and Sr
isotopic variation; nonetheless, it is
important to inquire whether the
pseudoisochrons do contain meaningful
age and isotopic information about the
mantle.
One means of evaluating pseudo-
isochrons is to determine whether they
involve identifiable mixing-line
chemistry. Rb-Sr variation diagrams
and plots of 87 Sr/ 8,i Sr versus 1/Sr can be
used for this purpose. Examination of
such plots for the data used to build
Table 6 reveals that a good pseudoiso-
chron is commonly accompanied by a
good positive correlation between
( 87 Sr/ 8,i Sr) and 1/Sr but no correlation
at all between Rb and Sr. The absence
of a Rb-Sr correlation is inconsistent
with simple magma mixing. Thus,
while we cannot discount mixing of
heterogeneous mantle material, the
data in general seem to preclude sim-
ple two-component magma mixing as
an important factor in most
pseudoisochrons. The volcanics of Peru
are among the few examples that dis-
play good correlations between ( 87 Sr/
8,; Sr) and 1/Sr, and Rb and Sr (James et
al., 1976). In this case, the pseudoiso-
chrons could result either from mixing
or selective melting of a heterogeneous
source region of uniform age or from
disequilibrium melting of homogene-
ous mantle.
Selective withdrawal of magma from
mantle material that is heterogeneous
(and of uniform or nonuniform age)
will result in the generation of
pseudoisochrons that do not necessar-
182
CARNEGIE INSTITUTION
720
CO
CO
715 -
710
705
u im
B
706
•
•
\^
0704
•
702
i i
i
i
002 0.04 006 0.1
706
c
•
0.704
•
•
J^i
• -
•
• ^^^
•
702
•
•
1 1
1
I
0.02 04 06 0i
02 04 06
720
716
712
708
704
05 0.10 015 20
Fig. 30. Rb-Sr mantle isochrons (initial 87 Sr/ 86 Sr versus Rb/Sr) for some representative igneous rock
associations of continental affinity (for more detail, see Table 6). (A) Quartz and olivine norites from
the Arnage (open circles) and Haddo House (closed circles) igneous masses, Scotland. (B) Alkali basalt
series of the Western Grand Canyon, U.S.A. (C) Mafic rocks of the Archean Usushwawa Complex,
South Africa. (D) Colorado Plateau basalts, U.S.A. (E) Snake River Basalts, U.S.A. Closed Circle,
Craters of the Moon basalts; Open circle, King Hill basalts. The field is for olivine tholeiites (N = 34)
with the cross showing the average composition. (F) Peruvian Cenozoic volcanics, Southern Peru.
Closed circle, Barroso volcanics; open circle, Arequipa volcanics.
DEPARTMENT OF TERRESTRIAL MAGNETISM
183
G
\\05i-
0.706
704
08
o
o • O- ^
•3>®
o
%
"~~~~ o
0702
1
1
1
0712
0710
0.10 020 0.30 040
708 -
0706
0.704-
0702
H
_
• •
- -
^•__
•
•
•
»^
-
i
i
i
i i
0.10 0.20 0.30 040 50
720
715
710
0.705 -,
700
(Rb/Sr!
Fig. 30. (G) New Zealand East Arc volcanics.
Closed circle, basalts; open circle, andesites. (H)
Mesozoic granites of California, U.S.A. (initial
* 7 Sr/" H Sr > 0.706). (I) Jurassic dolerites of Ant-
arctica (includes the Ferrar Group and Queen
Maud Land dolerites).
ily reflect the average age of the source
region; however, we cannot avoid the
conclusion that such pseudoisochrons
reflect a' long history of isolated
isotopic evolution of the subcontinental
mantle. Thus, whatever the specific
origin of pseudoisochrons, they do
imply a significant age for mantle
heterogeneities, and for this reason we
refer to them as mantle isochrons, a
general term designed to emphasize
the unique nature of the ancient
isotopic imprint borne to the surface in
mantle-derived magmas.
Accepting that pseudoisochrons are
mantle isochrons places major con-
straints on the nature of the mantle
source regions. First, the age informa-
tion embodied in mantle isochrons de-
mands the existence of mantle that has
long preserved a distinct chemical
identity. This in turn demands that the
mantle source region be sufficiently
immobile as to be static and resistant
to reworking and mixing processes dur-
ing most of its history. Second, the ele-
vated Sr-isotopic ratios indicated by
the mantle isochrons suggest that the
mantle source region is not chemically
depleted. Third, the wide range in ob-
served Sr-isotopic composition requires
that the mantle be grossly heteroge-
neous, probably on both regional and
local scales (Hofmann, 1975; Hofmann
and Hart, 1975).
What sort of mantle fits these re-
quirements? Asthenosphere most
probably does not, leaving either sub-
continental lithosphere or mesosphere
as the likely source region. The linear
arrangement of continental island arc
volcanoes associated with magmatic
184
CARNEGIE INSTITUTION
activity derived from above subducted
lithosphere would seem to preclude
mesosphere as a likely direct source,
even though its nature and composi-
tion are probably compatible with the
mantle isochrons. Is continental
lithosphere compatible with these re-
quirements, however? In the following
section we will show that there is a
substantial amount of evidence that al-
lows an independent case to be built for
thick, ancient, heterogeneous and
chemically undepleted continental
lithosphere with an isotopic composi-
tion consistent with that implied by
mantle isochrons.
Nature of continental lithosphere.
Most of our knowledge concerning the
nature of continental lithosphere
comes from three sources: seismic
studies, heat-flow measurements, and
the analysis of ultramafic nodules be-
lieved to be samples of lithosphere
rafted to the earth's accessible regions
in mantle-derived magmas. These
broad areas are so interdependent that
it is impractical to isolate the indi-
vidual contributions originating in
each. Consequently, in this discussion
we synthesize the evidence from these
sources by treating successively the
questions of thickness of continental
lithosphere, chemical composition of
lithosphere (and the resulting implica-
tions for heat flow), isotopic heteroge-
neity of lithosphere, and the age of
lithosphere.
Although there is no consensus
among geophysicists on the thickness
of continental lithosphere, with esti-
mates ranging between about 200 and
400 km (Dziewonski, 1971; Kanamori,
1970; Jordan, 1975; Okal and Ander-
son, 1975; Sipkin and Jordan, 1975;
Toksoz and Anderson, 1966), the lack
of agreement is largely one of defini-
tion and not substance. In particular,
the lower estimates of lithosphere
thickness are based on the determina-
tion of the minimum depth below
which oceanic and continental velocity
structures need not differ. But depths
so calculated provide no information on
actual lithosphere thickness; that
thickness is determined by the depth at
which shear-wave velocities or Q de-
crease rapidly or discontinuously to as-
thenospheric values. By this criterion,
few seismic studies successfully iden-
tify the base of the lithosphere. They do
show, however, that lithosphere be-
neath cratons may be at least 300-400
km thick (Alexander and Sherburne,
1973; Goncz and Cleary, 1976); that is,
the continental mantle is subsolidus to
that depth, exhibiting no significant
low velocity or low Q zone at lesser
depths. While there may be a low-
velocity channel at greater depths, the
poor resolving power of the seismic
data precludes its unequivocal identifi-
cation. One important study of Q be-
neath the Andes of South America,
however, clearly demonstrates the
existence of a low-rigidity (and proba-
bly low-velocity) channel at a depth of
350 km beneath a high-rigidity lithos-
phere (Sacks, 1969; Sacks and Okada,
1974). This result is of considerable
general importance because the An-
dean orogen evolved atop the ancient
South American shield, and the integ-
rity of the underlying lithosphere ap-
pears to be intact (James, 1971).
Additional evidence favoring thick
continental lithosphere comes from an
analysis of global heat-flow data by
Pollack and Chapman (Pollack and
Chapman, 1976; Chapman and Pol-
lack, 1975). Although the results of
their study are not conclusive, they in-
dicate that the lithosphere beneath an-
cient continental interiors may attain
thicknesses in excess of 300 km. This
estimate is based upon regional varia-
tion in geothermal gradient and the
corresponding depths at which the
geotherm intersects an empirically
postulated temperature of incipient
melting. Pollack and Chapman (1976)
take the temperature at the litho-
sphere-asthenosphere boundary to be
0.85 the temperature of dry melting of
peridotite. This value is obtained em-
pirically by considering the thickness
of oceanic lithosphere as a function of
varying oceanic geothermal gradients.
The general form of continental geo-
DEPARTMENT OF TERRESTRIAL MAGNETISM
185
thermal gradients inferred from heat
flow is corroborated by the results of
geochemical studies in which experi-
mentally determined phase relation-
ships and mineral compositions of
coexisting pyroxenes as functions of
temperature and pressure are applied
to nodules to obtain estimates of their
equilibration temperatures and
pressures (e.g., Boyd and Nixon, 1975).
The persistence of subsolidus tempera-
tures to great depth is further attested
to by the absence of interstitial glass in
ultramafic nodules (e.g., Kudo et al.,
1972). The Pollack and Chapman
model, as in most previous thermal
models (Clark and Ringwood, 1964;
Sclater and Francheteau, 1970), as-
sumes that the continental lithosphere
is somewhat depleted in heat-
producing elements. Yet, one of the
principal conclusions to be drawn from
the interpretation of mantle isochrons
is that continental lithosphere is essen-
tially undepleted. Since chemical com-
position is the input to the thermal
model, our conclusion must be consis-
tent with both the observed heat-flow
data and with what is known about the
composition of continental lithosphere
as revealed by studies of ultramafic
nodules found in kimberlites and
basalt (see Ringwood, 1975, pp. 100-
114 for general discussion and exten-
sive bibliography) .
Kimberlites crystallize from vola-
tile-rich magmas that appear to have
originated at depths greater than 200
km (Dawson, 1972; Boyd and Nixon,
1970; Ringwood and Lovering, 1970;
Ringwood, 1975). They apparently as-
cended rapidly, and in so doing incor-
porated many ultramafic xenoliths
which are believed to represent sam-
plings of different depth zones of the
mantle (Wagner, 1928; Dawson, 1972;
Harris et al., 1972; Harris and
Middlemost, 1969; MacGregor, 1975;
Boyd and Nixon, 1975; Lovering and
Tatsumoto, 1968). Although there is
debate about the source region of the
kimberlite magma and no assurance
that it lies in the lithosphere, there is
general consensus that many of the
xenoliths are subsolidus mantle frag-
ments showing little reequilibration to
lower temperature and pressure as-
semblages (MacGregor, 1968; Boyd and
Nixon, 1975; Dawson, 1972; O'Hara et
al., 1975). The nodule compositions can
therefore be used as guides to real
mantle compositions. Studies of
nodules in kimberlite indicate that the
subcontinental mantle has high Rb/Sr
ratios, correspondingly high 87 Sr/ 8<i Sr
ratios, and is otherwise undepleted in
its basaltic fraction (Rhodes and Daw-
son, 1975; Barrett, 1975; Ridley and
Dawson, 1975; Hutchinson and Daw-
son, 1970; O'Hara etal., 1975). The un-
depleted character of the mantle from
which the nodules were derived is at-
tested to by the dominance of the fertile
garnet peridotite (especially the
sheared variety) over other more de-
pleted nodules (O'Hara et al., 1975).
Thus, nodule studies corroborate our
conclusions based on Sr-isotopic com-
positions and argue for the existence of
a continental lithosphere that is far
from depleted and which, under favor-
able thermal conditions, would be a
ready source of basaltic magma.
Steady-state thermal models based
on undepleted mantle compositions can
be constructed that are entirely consis-
tent with measured heat flow and pos-
tulated geothermal gradients. Consid-
er, for example, a surface heat flux for
a continental craton of 1.05 heat flow
units (HFU), a standard 40-km crust
which produces 0.43 HFU, and a
lithospheric mantle 310 km thick that
consists of undepleted peridotite con-
taining 1000 ppm K, 0.06 ppm U, and
0.23 ppm Th. Crustal heat production
and steady-state heat flow are taken
from Sclater and Francheteau (1970),
and are consistent with the relation-
ship between heat flow and surface
heat production proposed by Roy et al.
(1968). K/U and U/Th ratios are taken
from Clark and Ringwood (1964). This
mantle material will contribute 0.38
HFU to the steady-state heat flux,
leaving 0.24 HFU to be derived from
below 350 km, a deep-mantle contri-
bution that is only about one half the
186
CARNEGIE INSTITUTION
value assumed in the static earth
model of Clark and Ringwood. Our
thermal model, in which most surface
heat flow originates in the lithosphere
rather than at greater depth, results in
slightly lower geothermal gradients for
the same surface heat flux; this in turn
forces the point of incipient melting to
even greater depths, thereby making
thick continental lithosphere even
more feasible. This example illustrates
that steady-state thermal models based
on undepleted mantle compositions are
entirely consistent with observed heat
flow and postulated geothermal gra-
dients.
The study of ultramafic nodules pro-
vides evidence not only that the conti-
nental lithosphere is chemically unde-
pleted but that it is ancient, isotopi-
cally enriched, and grossly heteroge-
neous. Kimberlite nodules typically
show 87 Sr/ 8( 'Sr ratios that vary between
0.703 and 0.710 (average, 0.707) (Bar-
rett, 1975; Allsoppeia/., 1969; Mitchell
and Crocket, 1971; Ridley and Dawson,
1975; Hutchinson and Dawson, 1970;
Harris et al., 1972). The consistent and
wide variation in isotopic ratio cannot
be attributed to contamination and
must reflect mantle inhomogeneity
(Barrett, 1975; Ridley and Dawson,
1975). Furthermore, the existence of
isotope disequilibrium between min-
eral phases in individual nodules
suggests that the heterogeneity is pre-
sent even on the scale of mineral grains
(Kudo et al, 1972; Paul, 1971; Com-
pston and Lovering, 1969; Steuber and
Ikramuddin, 1974; Burwell, 1975; Bar-
rett, 1975; Peterman et al, 1970; Har-
rises al, 1972; Dasch and Green, 1975).
The high average 87 Sr/ 8,i Sr ratio
suggests significant prehistory for con-
tinental lithosphere, long isolated from
less enriched asthenosphere. More di-
rect evidence that the subcontinental
mantle is very old comes from Sr and
Pb isotope studies on ultramafic
nodules. Measurement of apparent
ages indicates mantle ages up to 3 b.y.
or more (Burwell, 1975; Manton and
Tatsumoto, 1971; Paul, 1971; Steuber
and Ikramuddin, 1974; Dasch and
Green, 1975; Cooper and Green, 1969;
Harris et al, 1972). Other studies
showing wide variation in Sr isotopic
ratios between nodules provide indirect
evidence that parts of the mantle have
remained in isotopic isolation over very
long periods of time (Barrett, 1975;
Kudo et al., 1972; Compston and Lover-
ing, 1969). Thus while no single geo-
chemical study yet provides definitive
evidence of ancient mantle, the inte-
grated result of all studies argues for
an aged and heterogeneous subconti-
nental lithosphere.
Implications. Because continental
lithosphere exists at relatively low,
subsolidus temperatures, it is protected
from transport into subduction zones
by the buoyancy of the overlying conti-
nental capping; as a result, parts of the
lithosphere may be as old as the
coupled sialic crust (e.g., early Pre-
cambrian for continental cratons). Any
sections of lithosphere that are not de-
pleted will, by virtue of their sheltered
existence, possess distinctive isotopic
compositions that result from both re-
gional and local variation in Rb/Sr
ratios. If this ancient lithosphere par-
ticipates in younger magmatism, then
those volcanic products will contain
isotopic evidence of the source region's
"pre-history." This evidence, manifest
as mantle isochrons, raises two major
questions: first, how does this litho-
sphere form, and second, how does such
lithosphere, isolated for long periods of
time, become involved in young mag-
matism?
The range in initial 87 Sr/ 8(i Sr of
mantle-derived magmas (Table 6 and
Figure 30) provides strong evidence
that the continental lithosphere is
characterized by a degree of regional
isotopic heterogeneity that is more
pronounced than that observed in other
mantle provinces. Specifically, the sub-
continental mantle has Sr isotopic
ratios ranging from as low as 0.703 to
more than 0.710, a range of values
nearly identical to that found for
mantle-derived ultramafic nodules. By
comparison, ocean-ridge basalts, taken
to represent oceanic asthenosphere,
DEPARTMENT OF TERRESTRIAL MAGNETISM
187
display a quite limited range of values
(about 0.7027) (Hofmann and Hart,
1975), indicating that this astheno-
sphere is comparatively homogeneous
isotopically as might be expected in
rapidly convecting, partially molten
material. On the other hand, oceanic
islands, taken to represent at least part
mesosphere, exhibit a considerable
range in Sr-isotopic values (from about
0.703 to 0.706), which suggests that the
mesosphere has substantial regional
isotopic heterogeneities, although not
as pronounced as those of continental
lithosphere (Brooks et al., 1976; and
Sun and Hanson, 1975).
The upper limits of these ranges in
87g r /8«g r ra tios lead to an important in-
ference concerning the relative en-
richment of the different mantle pro-
vinces. Continental lithosphere consis-
tently has Sr-isotopic values that are
significantly higher than those of oce-
anic asthenosphere, indicating that
some parts of the subcontinental man-
tle are considerably more enriched in
radiogenic components and LIL (large-
ion-lithophile) elements than sub-
oceanic mantle. Mesosphere, as man-
ifest in oceanic islands, can be regarded
as a nondepleted base to oceanic and
continental regions alike, locally pos-
sessing a degree of enrichment in
radiogenic components comparable to
but possibly somewhat smaller than
that of continental lithosphere.
Our review of mantle isochrons and
related Sr-isotopic data indicates that
continental lithosphere is in part both
more enriched and more heterogeneous
in its radiogenic components (and LIL
elements) than other mantle provinces.
Such lithosphere enrichment and
heterogeneity could result from a vari-
ety of processes operating either indi-
vidually or in concert, and possible
models include:
(i) Thermal accretion by "freezing" of
enriched asthenosphere. Since litho-
sphere begins its formation soon after
crustal formation, the lithosphere
under old continental areas may have
started its formation at a time when
the asthenosphere was enriched or
nondepleted (as contrasted with its
present state of chemical depletion).
(ii) Trapping of an enriched as-
thenosphere boundary layer (formed by
progressive crystallization of the
lithosphere with concomitant concen-
tration of LIL elements into the
partial-melt zone of the asthenosphere
directly beneath the lithosphere boun-
dary).
(iii) Upward percolation of a LIL-
element and volatile-rich phase (not
miscible with the silicate partial melt)
from mesosphere or asthenosphere into
the subcontinental lithosphere.
(iv) Diapiric rise of LIL-element en-
riched mantle diapirs or blobs from the
mesosphere, which underplate and be-
come incorporated into the thickening
subcontinental lithosphere. The mech-
anism of rise of diapirs from a buoyant
mantle layer has been considered at
length by Ramberg (1972). The forma-
tion of oceanic island volcanism by as-
cent of chemically distinct blobs or
plumes forms a background to our
model and has been discussed by Schil-
ling (1975).
Model (i) would produce a zoned
lithosphere grading from enriched at
the top to depleted at the base (since
present-day asthenosphere is de-
pleted). Derivation of continental
magmas from such a lithosphere might
reflect this zoning, with magmas
formed at the greatest depths
(youngest lithosphere) carrying a de-
pleted signature. There is no convinc-
ing evidence, however, for the exis-
tence of significant quantities of "de-
pleted" magmas in continental regions,
and hence operation of this model re-
quires either that no magmas are de-
rived from this depleted zone of the
continental lithosphere, or that such
magmas reequilibrate with shallower,
more enriched lithosphere.
Model (ii) has been proposed by Kay
(1975) for the formation of het-
erogeneities in the suboceanic mantle;
however, consideration of the rates of
diffusion of LIL-elements in silicate
melts (Hofmann, 1975; Hofmann and
Hart, 1975), shows that any such en-
188
CARNEGIE INSTITUTION
riched boundary- me It layer will be
very thin (—10 m) and quantitatively
insignificant for the later production of
enriched magmas. The main difficulty
in this model (for both oceanic and con-
tinental regions) arises because litho-
sphere accretes at a velocity larger
than the scale lengths of LIL-element
diffusion in silicate melts, so that the
steady-state composition of the litho-
sphere would be close to that of the as-
thenosphere from which it forms, and
only trivial amounts of "enriched" par-
tial melt could result from this process.
Model (iii) requires a mobile phase
that is not miscible in the partial melt
phase of the asthenosphere at the
pressures and temperatures character-
istic of the asthenosphere. Such a
phase might be a hydrous, carbonate-
rich fluid that separates from the
mesosphere and rises buoyantly
through the asthenosphere to be trap-
ped against the advancing lithospheric
front. Later partial melting of litho-
sphere containing this enriched phase
could produce magmas showing en-
riched characteristics. In extreme
cases, this enriched phase might pro-
gress directly through the lithosphere,
forming members of the kimberlite-
carbonatite suite at the surface. While
this model has many plausible charac-
teristics (and is closely related to model
iv with which it could operate in con-
junction), we note that it also implies
the presence of kimberlite-carbonatite
magmas throughout the geologic rec-
ord. They are in fact rare in Precamb-
rian time, a time when presumably
thinner crust and lithosphere would
have greatly enhanced the likelihood of
such magmas reaching the surface.
Model (iv) involves a convecting
mesosphere containing thermal or
chemical heterogeneities (Anderson,
1975), such that blobs or plumes of ma-
terial can become separated, rise
through the asthenosphere and become
plastered-on and slowly incorporated
into the thickening subcontinental
lithosphere. This mesospheric material
is presumed to be enriched in the LIL-
elements relative to asthenosphere so
that the resulting subcontinental
lithosphere will be a heterogeneous
combination of frozen-out astheno-
sphere and included mesosphere-
derived blobs. This model is attractive
because it can be applied equally well
in oceanic regions to explain the exis-
tence of oceanic islands (variously re-
lated to hot spots or plumes) which in-
variably are built of volcanic material
enriched in LIL-elements and Sr and
Pb isotopes relative to the depleted,
asthenosphere-derived spreading-ridge
basalts.
In oceanic regions, the mesospheric
blobs or plumes (a plume may be sim-
ply a more or less continuous train of
blobs) do not necessarily go through a
two-stage process of lithosphere under-
plating and subsequent remelting. In-
stead, they may diapirically penetrate
the thinner oceanic lithosphere and di-
rectly supply the partial melt leading
to ocean island volcanism. If the blobs
do not penetrate the lower boundary of
oceanic lithosphere, they will produce
heterogeneity similar to that of the
continental lithosphere. By this con-
cept, then, oceanic regions provide us
with a recent (<200 m.y.) record of the
continuous upward migration of en-
riched blobs from the mesosphere, a
process which has been going on under
continents for billions of years.
The main difference between mantle
isochrons of oceanic and continental
regimes is that higher Rb/Sr and * 7 Sr/
8,i Sr ratios are found for continental
volcanics. This implies that there must
be an additional fractionation (in-
crease) of Rb/Sr ratio associated with
the two-stage continental case (under-
plating and subsequent remelting) re-
lative to the single-stage oceanic case
(direct penetration). We explain this as
follows: At some point in the rise of
mesospheric blobs, they become par-
tially molten; this silicate melt, with a
higher Rb/Sr than the initial value of
the blob, becomes partially segregated
DEPARTMENT OF TERRESTRIAL MAGNETISM
189
from the crystalline residue as the blob
ascends and is plated onto the bottom
of the continental lithosphere. Later
partial melting of this material will
favor the previously partially segre-
gated melt, leading to a second stage of
Rb/Sr enrichment. The relatively high
87 Sr/ 8( 'Sr ratios observed in continental
volcanics (relative to oceanic-island
volcanics) arise because of the isotopic
aging that follows the initial Rb/Sr
fractionation. Cases of more or less
complete melt-crystal segregation
prior to or during the underplating
stage would result, on further cooling,
in an eclogite solid (frozen melt) and
ultramafic residue. The eclogite nod-
ules commonly brought to the surface
in kimberlitic rocks may be evidence
for such a segregation process.
The combined weight of the evidence
discussed in this paper suggests that
continental lithosphere accretes
through a combination of processes (i)
and (iv). Such a model provides a satis-
factory explanation for mantle iso-
chrons from both continental and
oceanic regions. A schematic view of
this model is given in Fig. 31A, and
although it is recognized that some as-
pects of the other models discussed may
be of local importance, we have re-
stricted our figure to a portrayal of the
main features of continental litho-
spheric growth.
The data of Table 6 show that mantle
isochrons are associated with a number
of specific tectonic environments where
ancient lithosphere can become reacti-
vated and involved in partial melting
processes. These environments are
created where, (1) subcontinental
lithosphere overlying a subduction
zone is activated by proximity to that
zone; (2) continental-scale rifting forms
fractures that penetrate deep, un-
loaded lithosphere; and (3) a
continent-bearing plate over-rides an
oceanic rise or plume.
Schematic representation of these
reactivation environments is illus-
trated in Fig. 3 IB.
Recognition that ancient lithosphere
can play a major role in younger mag-
matism in these tectonic environments
allows us to formulate hitherto un-
treated models of magma genesis.
As an example of subduction zone
reactivation, consider the well-studied
Cenozoic volcanic rocks of the Andean
region of southern Peru. These provide
documentation of the role of ancient
continental lithosphere in producing
both high * 7 Sr/ Kli Sr ratios and well-
defined mantle isochrons in subduction
zone environments (James et at., 1976).
Here, the ancient lithosphere of the
South American shield extends be-
neath the Andes to a depth of at least
200 to 300 km and forms, throughout
the depth range over which magma is
generated, a continuous lithosphere-
lithosphere boundary with the de-
scending Nazca plate. The model which
most plausibly explains the isotopic
and trace element data is one in which
andesitic magma is produced by sub-
duction zone melting of the ancient
South American lithosphere. The
isotopic imprint of that aged litho-
spheric mantle material is preserved in
the mantle isochrons and anomalously
high 87 Sr/ 8,i Sr ratios of the young vol-
canics of the Andean region.
The rifting-related reactivation en-
vironment is represented by the alkalic
complexes of Africa and the Jurassic
tholeiitic flood basalts (and dolerite
sheets) associated with the break-up of
Gondwanaland. The latter define
mantle-isochrons both on a local (Ant-
arctica) and regional (Antarctica,
Tasmania, and Africa) scale (Table 6),
and document the existence of very un-
depleted mantle in these regions. Man-
tle isochrons are also common in re-
gions where a continent appears to
have over-ridden an ocean rise or man-
tle plume. Our examples all come from
western U.S.A., where many rock asso-
ciations have mantle isochron ages in
the range 600 to 1000 m.y. This par-
ticular region is probably the most
complex of those discussed, and despite
190
CARNEGIE INSTITUTION
LITHOSPHERE
ASTHENQSPHERE
KJ . Oj v .,.• MESOSPHERE
©
(D
Fig. 31. Diagram showing development (A) and subsequent reactivation (B) of blob-enriched
continental lithosphere (vertical exaggeration about 5 = 1). (A) Blobs are released from mesosphere
by buoyancy and convection, ascend through convecting asthenosphere, and penetrate through either
under-plate thickening continental lithosphere or oceanic lithosphere. Ascending blobs (diagonally
ruled pattern) may be partially molten, with some segregation of liquid occurring before they become
subsolidus in the lithosphere (solid pattern). Dashed "age" lines in lithosphere represent progressive
positions of the boundary representing accretion of lithosphere from asthenosphere. Note that oceanic
lithosphere is shown to be younger than 200 m.y. An oceanic "plume" is indicated schematically as a
continuous train of blobs. (B) Various reactivation environments, where partial melts (stippled pat-
tern) are formed in continental lithosphere: (1) above a subducting oceanic plate; (2) over a thermal or
upwelling anomaly in the asthenosphere (plume or overridden ridge, arrows); and (3) during conti-
nental rifting (initial stage shown without causal mechanism).
DEPARTMENT OF TERRESTRIAL MAGNETISM
191
numerous studies a definite pattern of
ancient lithosphere consumption is not
yet forthcoming. However, the possible
participation of the lithosphere in the
magmatism of the region is discussed
in the literature (Hedge and Noble,
1971; Leeman, 1974, 1975; Leeman
and Manton, 1971; Mark et al, 1975;
Petermane^a/., 1970).
An important aspect of Table 6 is
that mantle isochrons are not re-
stricted to young volcanics but are also
encountered in young hypabyssal and
plutonic intrusives (Jurassic dolerites
of Antarctica; California Mesozoic
batholiths), metamorphosed basic
rocks of early Palaeozoic age (Caledo-
nian basic-igneous Province, Scotland),
and in ancient layered-intrusions
(Usushwana complex of South Africa).
These examples are encouraging be-
cause they demonstrate that mantle
isochrons are not necessarily masked
by either differentiation or meta-
morphism, thereby considerably ex-
tending their usefulness as petrogene-
tic indicators.
One consequence of the mantle iso-
chron model is that crystallization ages
determined on basic igneous rocks by
the Rb-Sr whole-rock technique can be
in excess of the true age by many hun-
dreds of millions of years. This problem
of inherited age is more serious for
younger rocks, and there are well-
documented instances of conflicts be-
tween stratigraphic age and Rb-Sr age
in the literature (e.g., Compston, 1974;
Cormier, 1969).
Effective use of the mantle isochron
concept requires knowledge of actual
crystallization ages (so that the meas-
ured isochron can be divided into its
pre- and postcrystallization compo-
nents), and determination of isochron
parameters on rocks that have been
subjected to minimal postmelting pro-
cesses (fractional crystallization, wall-
rock contamination, etc.). The concept
of Rb-Sr mantle isochrons as we have
presented it provides a new tool for un-
derstanding the petrogenesis of
mantle-derived igneous rocks.
Acknowledgments
This study has been supported in
part by the National Science Founda-
tion (Grant No. GA 36094) and the Na-
tional Research Council of Canada
(Grant No. A 5581 to C.B.). One of us
(C.B.) gratefully acknowledges the
Carnegie Institution of Washington for
providing financial support and re-
search facilities during his sabbatical
year.
PART II: OCEANIC REGIONS
Introduction
Tatsumoto (1966) showed that
basalts from oceanic areas preserved a
correlation between U/Pb ratio and
lead isotopic ratio, and concluded that
this correlation, presumably inherited
from the mantle source of the basalts,
suggests a mantle in which U/Pb
heterogeneities have existed for a bil-
lion years or more. A similar general
correlation was noted between * 7 Sr/ H,i Sr
and Rb/Sr for oceanic basalts (Hart,
1971; Allegre, 1972). Recently, this
systematic Rb-Sr isochron correlation
was emphasized by Sun and Hanson
(1975), who showed that Sr isotope
data for alkali basalts (and nepheli-
nites) from 14 oceanic islands defined a
correlated array which, if interpreted
as an isochron, suggested an age for
mantle heterogeneities of about 2000
m.y. We were impressed with these re-
lationships and undertook a careful
screening of published data to look at
the Rb-Sr systematics of both tholeiitic
and alkali basalts from oceanic re-
gions. We found striking regularities
in almost all cases and present these
results because of their importance in
constraining the space and time evolu-
tion of mantle chemistry.
Compilation of Data
The data were compiled from the lit-
erature, using only those analyses of
relatively high precision which could
192
CARNEGIE INSTITUTION
be confidently normalized to an Eimer
and Amend standard value of 0.7080.
For some islands the published 87 Sr/
8, 'Sr was not accompanied by Rb/Sr
data. In these cases, powders of the
original sample were requested (and
are here acknowledged with thanks),
and the Rb/Sr ratio was measured
using conventional x-ray fluorescence
techniques. In other cases, unpublished
data were made directly available to
us, and these are also gratefully ac-
knowledged.
During the oceanic-island compila-
tion, we attempted to retrieve data for
basaltic rocks only, because dif-
ferentiation effects might distort the
Rb-Sr systematics. This screening, to-
gether with the identification of asso-
ciation (tholeiitic or alkali basalt) was
based on published assignments; for
some islands earlier literature was
searched for chemical analyses and de-
scriptions as an aid to proper evalua-
tion. The rock types from some islands
were classed as transitional, possessing
chemical compositions between
tholeiite and alkali basalt. Such rocks
were grouped with the alkali basalts in
order to keep the potentially "primi-
tive" tholeiitic association as unclut-
tered as possible.
Results and Treatment
* 7 Sr/*' ; Sr- Rb/Sr analyses were re-
trieved for 167 alkali basalts and 109
tholeiites. There are three principal
ways in which correlations between
Rb/Sr and H7 Sr/ 8,1 Sr can be evaluated:
(1) On a local scale using individual
data. Wherever the number of samples
permits, intra-island (or island group)
relationships can then be examined. (2)
On a regional scale using individual
data. The islands from a given ocean
can be grouped and the relationships
between the individual data examined.
(3) On a regional scale using averaged
data. The data for each island (or is-
land group) can be averaged to give a
single representative data point, and
the relationships between these aver-
age data points examined. Obviously
treatments (2) and (3) will provide the
same result only if each island is repre-
sented by a comparable number of in-
dividual analyses. However, many is-
lands are represented by a few
analyses, and a few islands by many
analyses, and it is the latter that then
dominate the relationships obtained by
use of treatment (2). This is well por-
trayed in a trial comparison of the
Pacific and Indian Ocean tholeiites.
The averaged island data are well cor-
related (r = 0.90) and form an 1100-
m.y. isochron. The individual data
however are uncorrelated (r = 0.12)
and do not define an isochron. In this
example, the difference is due to the
sample weight of the Hawaiian Islands
(n = 30), an island group that displays
no internal correlation between Rb/Sr
and K7 Sr/ 8H Sr on even an intra-island
scale. The manner in which the data
are treated is essential to the discus-
sion that follows, and we are confident
that the average island treatment (3)
provides an insight to the regional
Rb-Sr systematics of mantle-derived
volcanics that otherwise would be lost
in the more conventional individual-
data treatments. Consequently we will
use only treatments (1) and (3) in our
appraisal of the data.
The averaged data for the oceanic is-
lands (or island groups) are given in
Table 8, together with the number of
samples and the appropriate source(s)
of the data. When plotted on an Rb/Sr
versus * 7 Sr/ K(i Sr (isochron) diagram
(Fig. 32) the averaged data show two
striking features: Tholeiites and alkali
basalts occupy distinct and essentially
nonoverlapping fields; and each field
indicates a pronounced positive corre-
lation between K7 Sr/ K,i Sr and Rb/Sr. The
tholeiite data appear more closely cor-
related than the alkali basalt data, as
there are four alkali-basalt data points
which appreciably extend that field
toward higher Rb/Sr ratios. We note
that two of these island points (Gough
and Rodriquez) contain only one and
two individual analyses, respectively,
DEPARTMENT OF TERRESTRIAL MAGNETISM
193
TABLE 8. Average Rb/Sr and K7 Sr/ Hli Sr for Basalts from Oceanic Islands
(or Island Groups) and the Ocean Floors
Graph
Data
Location
Rb/Sr
"Sr/ Wi Sr
N*
No.t
Sources^
Oceanic island tholeiites
Bouvet
0.016
0.70369
4
1
1
Hawaiian Islands
0.017
0.7038
30
2
2,3,4,5,6
Iceland
0.014
0.70307
13
3
1,7,(A)
Kerguelen
0.033
0.7048
5
4
8
Kolbeinsey
0.009
0.70290
1
5
1
Reunion
0.033
0.7042
8
6
8,9,(B)
Samoa
0.049
0.7053
5
7
10,11
St. Pauls
0.039
0.7045
7
8
8,12
Ocean floor tholeiites
Indian
0.015
0.70314
8
9
13,14
Pacific
0.011
0.70265
15
10
13,15
Atlantic
0.008
0.70264
13
11
13,(A)
Oceanic island alkali basalts
Amsterdam
0.045
0.7039
4
12
8
Ascension
0.066
0.7028
4
13
1,16
Azores
0.046
0.7035
26
14
1,17,18
Canary
0.032
0.7033
7
15
1,5
Cape Verdes
0.028
0.7031
4
16
19
Crozet
0.060
0.7041
2
17
8
Easter
0.019
0.7030
7
18
20,(C)
Eniwetok
0.048
0.7038
2
19
21,(C)
Gough
0.059
0.7033
1
20
16
Guadeloupe
0.024
0.7033
4
21
20,(C)
Hawaiian Islands
0.028
0.7033
32
22
4,5,6,(D)
Iceland
0.040
0.7032
10
23
1,7,(A)
Jan Mayen
0.074
0.7036
9
24
1,22
Kerguelen
0.069
0.7052
4
25
8
Madiera
0.036
0.7031
1
26
5
Rodriguez
0.077
0.7038
2
27
9,(B)
Samoa
0.067
0.7054
14
28
11,23
St. Helena
0.034
0.7030
18
29
10,24
Tahiti
0.047
0.7042
9
30
20,(C)
Tristan de Cunha
0.062
0.7052
7
31
1,5
*N = number of samples averag
t Reference number for figures.
tSee Table 9.
and that the Jan Mayen point may be
anomalous because of the possibility
that the island is built on continental
crust.
When individual samples are
examined, only two islands, Kerguelen
and Samoa, show an intraisland corre-
lation with a positive slope (Fig. 33).
The other islands show either horizon-
tal data arrays (constant H7 Sr/ 8li Sr,
variable Rb/Sr, e.g. Iceland, Tristan da
Cunha, the Azores), equidimensional
fields with significant variation in both
directions (e.g., St. Helena, Hawaii), or
correlations with a negative slope (Ca-
nary Islands, Jan Mayen, Cape Verde
Islands). We find it interesting that the
only two islands which internally show
positively correlated data are the two
which have the highest average K7 Sr/
H,i Sr (and Rb/Sr) and we shall return to
this point.
194
CARNEGIE INSTITUTION
TABLE 9. Reference Key to Table 8
Key No.
Reference
1 O'Nions and Pankhurst ( 1974)
2 Hart (1973)
3 Heier, Compston, and McDougall
(1965)
4 Powel 1 and DeLong ( 1 966 )
5 Bence(1966)
6 Lessing and Catanzaro ( 1964)
7 Hart, Schilling, and Powell ( 1973)
8 Hedge, Watkins, Hildrett, and
Doering(1973)
9 McDougall and Compston ( 1975)
10 Hedge ( 1966)
11 Hedge, Peterman, and Dickinson
(1972)
12 Girod, Camus, and Vialeth (1971)
13 Hart (1976)
14 Subbarao and Hedge ( 1 973 )
15 Subbarao (1972)
16 Gast, Tilton, and Hedge ( 1964)
17 White, Hart, and Schilling ( 1975)
18 Hart and Brooks ( 1974)
19 Klerkx, Deutsch, and DePaepe ( 1974)
20 Peterman and Hedge ( 1972)
21 Hedge and Peterman (1970)
22 Lussia, Berdou, Polve, and
Vidal(1973)
23 Hubbard (1971)
24 Grant, Powell, Walther, and
Burkholder(1976)
(A) S. R. Hart (unpublished data)
(B) C. Brooks (unpublished data)
(C) C.E. Hedge (unpublished data)
(D) M. A. Lanphere, G. B. Dalrymple,
and D. A. Clague (unpublished data)
To decide if the positive correlations
shown in Figures 32 and 33 are statis-
tically significant, the data of Table 8
have been appraised by calculating the
correlation coefficient r, and testing
the slope of the fitted line for signifi-
cance against zero. The regression
treatment used to obtain slopes and in-
tercepts was that of York (1966); the
errors in the "age" and initial ratio are
calculated solely from the data scatter
and are given at 1 standard deviation
(following recommendations of Brooks
et al., 1972). While it is convenient to
express the slopes of these correlated
arrays as ages, we emphasize here that
these "ages" need not have time sig-
nificance. This point is discussed in
considerable detail later in the paper.
The statistical parameters relevant
to oceanic islands are presented in
Table 10. The significance levels of the
correlation coefficients range from 95%
(e.g., for Samoa and Kerguelen) to
greater than 99.99% (e.g., for the com-
plete tholeiite suite including ridge
basalts, and for the high-density por-
tion of the alkali-basalt field). We find
these correlations remarkable and
suggest that these data provide impor-
tant and hitherto unused constraints
relating to processes of oceanic mag-
matism.
Discussion
Possible causes for the positive iso-
chron correlations presented in the
previous section include surficial wea-
thering and contamination by crust,
sediments, or sea water; and inher-
itance of these isochron parameters
from a heterogeneous and aged mantle.
We believe direct weathering is
ruled out because of the fresh and unal-
tered nature of most of the samples in-
volved. Other contamination processes
might include mixing of low * 7 Sr/* ,; Sr
magma of direct mantle derivation
(ridge basalts) with more radiogenic
material represented by sediments or
oceanic crust which has interacted
with sea water. Oceanic sediments may
have * 7 Sr/ K,i Sr values as high as 0.717
and Rb/Sr values greater than 1.0
(Dasch, 1969; Biscay and Dasch, 1971;
Church, 1971). Only limited data exist
for altered oceanic crust, but material
has been analyzed with Rb/Sr ratios of
0.15-0.25 (Dasch et al., 1973; Hart,
1969), and presumably * 7 Sr/ H,; Sr ratios
could rise to values as high as sea
water (0.709). Thus, both sediments
and altered oceanic crust could provide
a mixing component that would pro-
duce positive correlations of the kind
shown in Fig. 33. This mechanism has
been described in detail in earlier pa-
pers and generally dismissed as a sig-
nificant effect in the petrogenesis of
oceanic basalts (Tatsumoto, 1966;
Powell and DeLong, 1966; Gast et al.,
1964; Grant et al, 1976). The fact that
DEPARTMENT OF TERRESTRIAL MAGNETISM
195
CO
CD
CO
00
CO
CD
^>^
o
CD
Samoa
i
i i
0.02 0.04 0.06 0.08 0.10 0.12
(Rb/Sr)
Fig. 32. K7 Sr/ K,i Sr- Rb/Sr isochron plot of tholeiites (filled circles) and alkali basalts (open circles)
from oceanic islands. Each point represents an average of available data on individual samples from a
given island or island group. Numbers refer to the data index in Table 8.
the lead isotope regression line for
oceanic basalts is distinctly different
from the regression line for oceanic
sediment (and sea water) also argues
against such a contamination mecha-
nism (Sun and Hanson, 1975; Chow
and Patterson, 1962; Church and Til-
ton, 1973; Meijer, 1976). Furthermore,
the marked uniformity of H7 Sr/ 8li Sr
values for some islands (O'Nions and
Pankhurst, 1974; Hart et al., 1973),
and the general lack of a positive iso-
chron (mixing-line) correlation be-
tween individual intra-island samples
also argues against a sediment-crust
contamination model. As we have dis-
cussed, only Samoa and Kerguelen
show intra-island "isochrons," whereas
these might be expected to be ubiquit-
ous under the operation of a contami-
nation model. Some islands do show
correlations of negative slope, as would
be expected from contamination pro-
cesses involving carbonate.
The conventional interpretation of
oceanic island strontium (and lead)
0.706
-
0.705
•
^•s--^
o--""^
0.704
n7fH
i
•
i
Kerguelen
i i i i
0.02 0.04 0.06 0.C
0.10 012
(Rb/Sr)
Fig. 33. Isochron plot of data for individual
samples of tholeiite (filled circles) and alkali
basalt (open circles) from the Samoa Islands and
Kerguelen Island. Best-fit regression lines and
slope values are shown.
196
CARNEGIE INSTITUTION
TABLE 10. Mantle Isochrons of Rb/Sr versus Present-Day
87 Sr/ 8H Sr for Basalts of Oceanic Regions
Location Identity
Age (m.y.)*
( K7 Sr/ 8,i Sr)*
Nt
Oceanic Islands
Alkali basalts: all averaged island data
Alkali basalts: all data except points
3,20,24 and 27
Tholeiites: all averaged island data
Tholeiites: all averaged island data
plus MORB§
Kerguelen, tholeiites: individual data
Kerguelen, alkali basalts: individual data
Kerguelen, all basalts: individual data
Samoa, tholeiites: individual data
Samoa, alkali basalts: individual data
Samoa, all basalts: individual data
600±210
0.7025±4
20
0.56(3)
1190±150
0.7017±3
16
0.91(4)
1430 ±180
0.7025±2
8
0.96(4)
1570±150
0.7023±2
11
0.97(4)
485±210
0.7041±4
5
0.80(1)
248 ±20
0.7045±1
4
0.99(2)
355±90
0.7042±2
9
0.83(2)
466±(250)
0.7044±3
5
0.74(1)
584±200
0.7039±6
14
0.65(1)
510±150
0.7041±4
19
0.62(2)
* Errors given at lcr
+ Number of samples regressed.
t Pearson correlation coefficient. Value in parentheses indicates at what level of confidence the
fitted regression line differs from zero (given in increments of standard deviation).
§MORB indicates mid-ocean ridge basalts.
isotope variations ascribes these to de-
rivation from an isotopically heteroge-
neous mantle (a review of this model is
given by Hofmann and Hart, Year
Book 74, p. 195, 1975). We consider this
the most tenable model at present, and
advance it here to explain the "iso-
chron" correlations shown in Fig. 32.
We view these positive "isochron" cor-
relations as being inherited by oceanic
basalt during derivation from a heter-
ogeneous suboceanic mantle.
Rb-Sr coherence during basalt
genesis. Figure 32 shows that oceanic
island tholeiites and alkali basalts
have a similar range of 87 Sr/ 8fi Sr values
and suggests that the separation of
data fields is due to higher Rb/Sr ratios
in the alkali basalts. On the average,
the alkali basalts of Table 8 have Rb/Sr
ratios about a factor of 2 higher than
the tholeiites (whereas the average
tholeiite 87 Sr/ 8fi Sr, 0.7037, is essen-
tially identical with the average alkali
basalt 87 Sr/ 8fi Sr, 0.7038).
We believe this is a significant ob-
servation and propose that the ob-
served Rb/Sr and strontium isotope
ratios of the tholeiites are directly rep-
resentative of the mantle source but
that the alkali basalts are displaced
toward higher Rb/Sr ratios relative to
the mantle source. The following dis-
cussion is an attempt to justify this
proposition.
While it is generally accepted that a
basalt will inherit the strontium
isotopic composition from its source
(Hofmann and Hart, Year Book 74, pp.
195-210), it is less obvious that a
basalt will also preserve the Rb/Sr
ratio of that source (because of the ef-
fect of partial melting and fractional
crystallization processes). This will
happen only if the partial melting pro-
cess leaves no residual mantle phase
which is a significant carrier of Sr and
Rb, and if all fractional crystallization
occurring after melting involves only
phases that are poor in Rb and Sr or
that do not fractionate Rb relative to
Sr. In the case of tholeiites, which in-
volve large degrees of melting, phases
that carry Rb or Sr (such as phlogopite
and amphibole) will probably not per-
sist after melting, so we would expect
the partial melt to have the same Rb/Sr
ratio as the original source material.
Phlogopite has been shown to have a
field of stability well above the wet sol-
idus of peridotite (Bravo and O'Hara,
1975; Modreski and Boettcher 1973;
Forbes and Flower, 1974; Boettcher et
al., 1975; Holloway and Eggler, this
DEPARTMENT OF TERRESTRIAL MAGNETISM
197
Year Book). However, the phlogopite
content of the mantle must be quite
small (probably less than 1% : Hofmann
and Hart, Year Book 74, pp. 195-210),
and phlogopite will be consumed dur-
ing a large degree of melting even if its
upper stability limit exceeds the tem-
perature at which melting occurs.
Amphibole may also be stable during
hydrous melting of peridotite
(Boettcher et al., 1975; Allen and
Boettcher, 1973). However, it is less ef-
fective than phlogopite at fractionating
Rb/Sr during large degrees of melting.
For example, for 25% melting of am-
phibole peridotite, where one half of an
original 20% of amphibole persists
after melting, the Rb/Sr ratio of the
liquid will be only slightly higher (6% )
than that of the starting material (us-
ing partition coefficients from Phil-
potts and Schnetzler, 1970). For
clinopyroxene, the partition coeffi-
cients of Rb and Sr are sufficiently
small (D Sr -0.08, D Rb -0.003) (Hart
and Brooks, 1974; Shimizu, 1974), that
residual clinopyroxene produces neg-
ligible (<3%) fractionation of Rb/Sr
during melting. Arguments similar to
these led Gast (1968) to the conclusion
that tholeiites formed by large degrees
of melting (15-25%) would contain
most of the Rb and Sr (and many other
trace elements) of the original mantle
source.
Gast (1968) also showed that frac-
tional crystallization is relatively inef-
fective in fractionating Rb from Sr.
Only crystallization of plagioclase will
be effective in changing the Rb/Sr
ratio, as the partition coefficients of Rb
and Sr in olivine and pyroxene are too
low to cause any significant fractiona-
tion of Rb from Sr. However, limits can
be placed on the extent of plagioclase
fractionation by noting that most
oceanic basalts have little or no detect-
able europium anomaly (Schilling,
1971; Sun and Hanson, 1975b). A de-
tectable europium anomaly (5-10%)
will be produced by crystallization of
15-25% of plagioclase, which in turn
will generate an increase in the Rb/Sr
ratio of the liquid of from 25-60% (us-
ing partition coefficients appropriate to
a 75-80% plagioclase (Philpotts and
Schnetzler 1970). We conclude that in
the case of oceanic tholeiites, the ex-
truded basalt very closely represents
the 87 Sr/ 8,i Sr and Rb/Sr ratios of the
mantle source.
Alkali basalts, on the other hand,
which are thought to be derived by rel-
atively small degrees of partial melting
(Gast, 1968; Green, 1973), may offer
more possibility of leaving behind re-
sidual phases such as phlogopite (Bes-
wick, 1976) or amphibole. In these
cases, using reasonable values for par-
tition coefficients (Philpotts and
Schnetzler, 1970) and phase propor-
tions, major changes in Rb/Sr ratios
can be brought about. For example, if a
3% melt fraction is derived from a
source containing 15% amphibole and
10% clinopyroxene and the amphibole
provides all of the melt (i.e., 12% am-
phibole and 10% cpx remain after melt-
ing), the Rb/Sr ratio of the melt will be
35-65% higher than that of the origi-
nal source (the range is due to selection
of two different sets of partition coeffi-
cients from Philpotts and Schnetzler
(1970). Phlogopite is surprisingly inef-
fective at fractionating Rb/Sr at low
degrees of melting because of the com-
peting effect of clinopyroxene (residual
clinopyroxene raises the melt Rb/Sr,
whereas phlogopite lowers it). A 3%
melt fraction of a source containing 1%
phlogopite and 20% clinopyroxene
(where 0.5% phlogopite and 2.5%
clinopyroxene enter the melt) will have
a Rb/Sr ratio about 10% higher than
the source material. Note that liquids
derived from a mantle source contain-
ing only phlogopite as a Rb-Sr carrier
will have Rb/Sr ratios lower than the
source.
These model calculations are not in-
tended to include all possible varia-
tions of parameters but only to suggest
approximate magnitudes of certain
limiting cases. To explain the higher
Rb/Sr ratio in alkali basalts than in
tholeiites (Fig. 32) we suggest that am-
198
CARNEGIE INSTITUTION
phibole, not phlogopite, is a necessary
residual phase during alkali basalt
genesis. For melts derived from depths
greater than the stability limit of ara-
phibole, reequilibration between such
melts and shallower amphibole-
bearing mantle may be required.
Alternative explanations for the
Rb/Sr and Sr isotopic pattern in
tholeiites and alkali basalts. In the pre-
ceding section, we gave our preferred
interpretation of the pattern observed
in Fig. 32. Some other conceivable ex-
planations are: (i) Alkali basalts and
tholeiites come from chemically dif-
ferent mantle sources, (ii) Alkali
basalts have undergone more frac-
tional crystallization of plagioclase (or
clinopyroxene) than tholeiites. (iii)
Basalt derivation from the mantle does
not occur under conditions of chemical
equilibrium between melt and residue
(O'Nions and Pankhurst, 1974). (iv)
Tholeiites and alkali basalts are gen-
erated sequentially by fractional melt-
ing from the same source material
(Beswick, 1976).
Explanation (i) is difficult to recon-
cile with the fact that many islands
erupt both tholeiites and alkali basalts
and that the strontium isotope ratios of
both are generally very similar. Com-
pare in Table 8, for example, the
isotope ratios of tholeiites and alkali
basalts from Iceland (0.7031 vs.
0.7032), Samoa (0.7053 vs. 0.7054),
Kerguelen (0.7048 vs. 0.7052), and the
Hawaiian Islands (0.7038 vs. 0.7033).
Explanation (ii) would require about
30% more plagioclase crystallization in
alkali basalts than in tholeiites.
Plagioclase crystallization of 30%
would lead to a consistent and easily
observable europium anomaly of some
15-20% alkali basalts relative to
tholeiites (presuming oxygen fugac-
ities and resultant REE partition coef-
ficients to be the same in both cases). In
addition, there should be a significant
"plagioclase" effect in the major-
element data and norms of alkali
basalts relative to tholeiites, and this is
also not observed (CaO and A1,0, con-
tents of oceanic tholeiites and alkali
basalts are quite similar). Extreme and
unrealistic amounts of clinopyroxene
crystallization would be required to
produce significant changes in Rb/Sr
ratios; the ratio is increased only 50%
after 99% crystallization.
Explanation (iii) is not consistent
with the relative coherence of trace
element patterns and the observed con-
stancy of major element composition in
various basalt types (Yoder, 1976).
Disequilibrium melting has also been
considered on the basis of existent dif-
fusion data (Hofmann and Hart, Year
Book 74, pp. 195-210), and judged to be
highly unlikely.
Explanation (iv), as proposed by
Beswick (1976), depends on the persis-
tence of a minor phase, phlogopite,
through a first stage of extensive par-
tial melting (tholeiite production). By
retaining K and Rb in residual phlogo-
pite, this first melt would have a lower
Rb/Sr (and higher K/Rb) ratio than the
bulk source. The reserve of K and Rb
stored in phlogopite is then released
during alkali basalt production, giving
rise to a higher Rb/Sr (and lower K/Rb)
ratio in the melt. As noted above, the
persistence of a minor refractory phase
is difficult to reconcile with an equilib-
rium involving large degrees of melt-
ing. No one, to our knowledge, has yet
studied a peridotite system containing
1% phlogopite or less, where residual
phlogopite remains after, say, 20% of
partial melting. Beswick's model also
requires a considerably higher K/Rb
partition coefficient for phlogopite-
melt (~3) than has yet been reported
(Philpotts and Schnetzler, 1970).
In summary, the alternative expla-
nations outlined in this section all ap-
pear unsatisfactory, and we conclude
that small degrees of melting with am-
phibole as a residual phase during al-
kali basalt genesis is the best explana-
tion for the differences in Rb-Sr be-
tween tholeiites and alkali basalt.
Comparison of Rb-Sr and U-Pb sys-
tematics. First, while tholeiites have
the same range of 87 Sr/* (; Sr as alkali
DEPARTMENT OF TERRESTRIAL MAGNETISM
199
basalts, it appears that tholeiites show
a smaller range in 206 Pb/ 204 Pb than al-
kali basalts (Tatsumoto, 1966; Sun and
Hanson, 1975a,b; Church and Tat-
sumoto, 1975; Sun and Jahn, 1975).
This may in part reflect the fact that no
lead isotope data are available for
tholeiites from the high 87 Sr/ 8(i Sr is-
lands such as Samoa and Kerguelen.
Second, the alkali basalts invariably
have higher U/Pb ratios than the
tholeiites. Furthermore, the observed
U/Pb ratios of alkali basalts are higher
than the inferred U/Pb ratios of the
mantle source (these can be calculated
from the lead isotope ratios), whereas
the observed U/Pb ratios of tholeiites
are similar to the inferred U/Pb ratio of
the source (Tatsumoto, 1966; Allegre,
1969). Thus the lead isotope syste ma-
tics provide strong evidence for a frac-
tionation of U relative to Pb during the
derivation process of alkali basalts,
and this supports the arguments pre-
sented above for an increase in Rb/Sr
ratio during alkali basalt genesis.
Finally, for many islands the lead
isotope ratios appear to exhibit signifi-
cant variations, whereas the strontium
isotope ratios appear almost constant.
For example, Iceland shows a 4%
spread in 2<,,i Pb/ 204 Pb ratio (Sun and
Hanson, 1975a; Sun and Jahn, 1975),
but only a barely detectable variation
in 87 Sr/ 8(i Sr (O'Nions and Pankhurst,
1974; Harte^ al., 1973). This effect may
be largely one of experimental preci-
sion in the isotope measurements. The
spread in Pb ratios on Iceland requires
a 35% range in source U/Pb ratio (av-
eraged over 1.5 billion years); a 35%
range in Rb/Sr ratio would produce
over the same time interval a spread of
87 Sr/ 8,i Sr of only 0.7029-0.7031. This is
close to the observed spread of Sr
isotope ratios for young Icelandic
tholeiites but is only barely significant
in terms of analytical precision.
Rb-Sr mantle isochrons. If the
tholeiite correlation of Fig. 32 is inher-
ited from a mantle source, what mean-
ing can be attached to the parameters
of this correlation? One model would
invoke a mantle that is heterogeneous
on a scale larger than the scale sam-
pled by individual island magmatism,
with the magmas essentially sampling
a continuum of mantle Rb/Sr systems.
The correlation of Fig. 32 thus would
be in a sense a mantle isochron, and the
slope could have definite age signifi-
cance. If the mantle heterogeneities
developed at a discrete time (two-stage
process), then the slope of the tholeiite
correlation shown in Fig. 32 suggests
an age of about 1.6 ±0.2 billion years
for this process. If the development of
the Rb-Sr heterogeneity was a continu-
ous or multiepisodic process, then any
"age" inferred from the slope is simply
some average age for the process as a
whole (it is possible to construct math-
ematical models of such processes in
such manner that subsequent mixing
of the appropriate reservoirs will give
rise to linear trends or "isochrons"
[Church and Tatsumoto, 1975]).
Another model would consider a
mantle that has only two discrete com-
positions, one at each end of the
tholeiite trend shown in Fig. 32, with
intermediate points representing vari-
ous mixtures of the two compositions.
For example, the mantle that supplies
MORB (asthenosphere) is relatively
uniform and distinct isotopically (Hof-
mann and Hart, 1975), and could be the
lower end-member. A uniform meso-
sphere might be the upper end-
member, with mixing of these two tak-
ing place during mantle convection and
diapirism. In this case, the slope of the
mixing line can only be interpreted di-
rectly as an age if the system from
which the two end-members developed
had a uniform strontium isotopic com-
position. A simple test for such a bi-
nary mixing process might be a linear
relationship between 87 Sr/ 8ti Sr and the
reciprocal of Sr concentration ( no such
linear relation is apparent in our data).
However, such a test is valid only if it
is applied to mixtures of melts rather
than mixtures of mantle source mate-
rial. The Sr concentration of melt
extracted from a mixed mantle source
200
CARNEGIE INSTITUTION
will vary inversely with the degree of
melting, and any significant variation
in the degree of melting would perturb
the expected inverse mixing relation-
ship between isotopic composition and
concentration.
At this point, it is important to note
that many oceanic islands have lead
isotopic compositions that define linear
trends (Sun and Hanson, 1975a, b; Sun
and Jahn, 1975). These trends are sub-
ject to the same uncertainty of in-
terpretation as the Rb-Sr trend of Fig.
32. They may be secondary isochrons,
implying the age of a geological event,
or they may be mixing lines. Ages cal-
culated from the slopes of these lead
isotope trends are in remarkable
agreement with the Rb-Sr "age" of the
tholeiite trend in Fig. 32. For basalts
from sea mounts and spreading ridges
in the N.E. Pacific, the Pb isotope trend
gives an age of 1.75 ±0.1 b.y. (Church
and Tatsumoto, 1975). Alkali basalts
from oceanic islands fall along this
same trend, with a Pb isotope age of
about 2 b.y. (Sun and Hanson, 1975a).
We feel this general concordance of
U-Pb and Rb-Sr systematics is not for-
tuitous but is strong evidence against a
random mixing process — i.e., any mix-
ing process in which the end-members
started with uncorrelated isotopic pa-
rameters. Granted this point, then
there is little essential difference be-
tween binary (or multicomponent) mix-
ing models and a two-stage model with
a continuum of subsystems: both result
in trends that have definite age signifi-
cance. The difficulty lies in distin-
guishing these two-stage models from
continuous or multistage models in
which the trends no longer have a spe-
cific age interpretation.
At present, the simplest model seems
to be that of mixing between relatively
radiogenic mantle (mesosphere) and
unradiogenic mantle (asthenosphere),
with the isochrons for both Pb-Pb and
Rb-Sr systems having a distinct age
significance. A definitive test of these
models will depend on further refine-
ment of U-Pb and Rb-Sr data, compari-
son with other decay schemes (Lu-Hf or
Sm-Nd), and analysis of materials from
similar tectonic settings in the past
geologic record.
Scale of mantle heterogeneities. We
have previously indicated that two is-
land groups display significant positive
intra-island Rb-Sr isochron correla-
tions, and we would like to suggest that
they are also derived from aged hetero-
geneous mantle. Two facts then need
explanation: Why do only these two is-
land groups retain a "mantle isochron",
and why is the apparent age of these
isochrons younger (360-510 m.y.) than
that of the oceanic island ensemble as a
whole (1200-1600 m.y.) and younger
than both intra-island and inter-island
Pb-Pb isochrons? Our ideas on this are
admittedly speculative, but we suggest
that the heterogeneities in the mantle
exist on variable size scales, and that
the more extreme mantle compositions
(note that Samoa and Kerguelen are
both the most radiogenic end-members
of the trends in Fig. 32) may only exist
as smaller volumes. Magmatism in a
given island group can thus sample a
finite variety of compositions from a
mantle in which the heterogeneities
exist on a smaller scale. At the same
time, the limited size of these
heterogeneities may allow for some
isotopic reequilibration between the
various systems, and thus the "age"
which would be detected by magmatic
sampling would be in some sense par-
tially "reset". This reequilibration or
internal redistribution may take place
by severe shear during diapirism or
during the final magmatic stage, keep-
ing in mind that diffusive transport
distances for Sr in silicate melts can be
of the order of 50 meters per 10 m.y.
(Hofmann, 1975; see also Hofmann and
Magaritz, this report).
Rb-Sr model calculations show that
redistribution of Sr can lead to partial
rotation of isochrons, with relatively
little disturbance of the linear aspect of
the data array (Farhat, 1975). Pb-Pb
isochrons will not be rotated by this
mechanism, as internal redistribution
of Pb in these cases will simply move
subsystems up or down the existing
DEPARTMENT OF TERRESTRIAL MAGNETISM
201
isochron. An analogy could be drawn
here between the behavior of mineral
isochrons and whole-rock isochrons
during the metamorphism of an igne-
ous pluton (the mineral systems are
smaller than the whole-rock systems
and may undergo isotopic reequilibra-
tion leading to clockwise rotation of
Rb-Sr mineral isochrons and con-
vergence of points on mineral Pb-Pb
isochrons). Further evaluation of these
concepts can be made when Pb isotope
data become available from Samoa and
Kerguelen.
The mantle isochrons discussed here
are compatible with the model outlined
in the preceding companion report by
Brooks et al. for the development of a
heterogeneous subcontinental litho-
sphere by the incorporation or under-
plating of ascending plumes or blobs.
The main difference is that there is no
old lithosphere in oceanic regions, and
the blobs or partial melts derived from
mesophere penetrate the oceanic
lithosphere directly, thus generating
oceanic island magmatism.
Summary and Conclusions
Rb/Sr ratios and * 7 Sr/ 8li Sr ratios,
when averaged by ocean basin (for
mid-ocean ridge basalts) and by
oceanic island or island group, yield
significant positive correlations on an
Rb/Sr isochron diagram. The correla-
tion for tholeiitic basalts is especially
strong, and we interpret it as a mantle
isochron that may date a major event
(or events) of mantle differentiation,
such as the segregation of a chemically
isolated asthenosphere 1.5-2.0 b.y.
ago. Alkali basalts show greater scat-
ter and are displaced toward higher
Rb/Sr ratios relative to the tholeiites.
We attach no direct age significance to
the alkali-basalt correlation because
these basalts are derived by much
smaller degrees of partial melting than
are tholeiites, so that the probability of
significant fractionation between Rb
and Sr (and therefore change of Rb/Sr
ratio) is much greater.
The Rb-Sr mantle isochron for
tholeiites is identical in age (1.6±0.2
b.y.) to ages (so-called secondary iso-
chrons) of oceanic rocks based on the
isotopic composition of lead (1.8±0.1
b.y.). This is an important finding, as it
strengthens the conclusion that the
isochron ages represent real geological
events rather than the result of ran-
dom mixing processes of no age signifi-
cance.
Although we do not argue that the
data could not also be explained by a
continuous or multiepisodic dif-
ferentiation and remixing event, the
simpler, first-order explanation, if cor-
rect, may date an event of fundamental
importance in the earth's history that
is not easily recognized from the sur-
face geological record.
We conclude by stating that the ap-
parent coherence of strontium isotope
ratio and Rb/Sr ratio in basaltic mag-
mas lends strong support to the cur-
rently popular ideas of a mantle in
which chemical heterogeneities have
existed on a long time scale (billions of
years). This coherence can be a power-
ful tool in mapping mantle heteroge-
neities and in understanding their
time evolution.
Acknowledgments
We acknowledge the valuable criti-
cisms and suggested improvements of
D. J. Whitford and C. Allegre. We also
gratefully acknowledge the following
people for providing either data or
sample powders: W. Compsotn, P.
Pushkar, C. E. Hedge, M. P. Groton, J.
B. Gill and M. A. Lanphere. M. Strobel
provided invaluable assistance with
statistical theory and relevant com-
puter programs. This study has been
supported in part by the National Re-
search Council of Canada (Grant No.
5581toC.B.).
References
Alexander, S. S., and R. W. Sherburne,
Crust and upper mantle structure in con-
tinental South America and its role in
global tectonics, Comptes Rendus No. 18,
202
CARNEGIE INSTITUTION
Part 1, IUGG, p. 85, Intern. Assoc, of Seis.
and Phys. of the Earth's Interior, Pro-
grams of the Scientific Assembly, Lima,
Peru, Aug. 20-31, 1973.
Allegre, C. J., Comportement des systemes
U-Th-Pb dans le manteau superieur et
modele devolution de ce dernier au cours
des temps geologiques, Earth Planet. Sci.
Lett., 5, 261, 1969.
Allegre, C. J., Geochronologie isotopique et
developpement de la croute terrestre au
cours des temps geologiques, in Structure
et Dynamique de la Lithosphere, p. 373, C.
J. Allegre and M. Mattauer, eds., Paris,
Hermann, 1972.
Allen, J. C, and A. L. Boettcher, Phase re-
lations and the stability of amphiboles in
an olivine nephelinite at high pressures,
Trans. Am. Geophys. Union, 54, 481,
1973.
Allsopp, H. L., L. O. Nicolaysen, and P.
Hahn-Weinheimer, Rb/K ratios and Sr-
isotopic compositions of minerals in ec-
logitic and peridotitic rocks, Earth Planet.
Sci. Lett., 5, 231-244, 1969.
Anderson, D. L., Chemical plumes in the
mantle, Geol. Soc. Am. Bull., 86, 1593-
1600, 1975.
Armstrong, R. L., A model for the evolution
of strontium and lead isotopes in a
dynamic earth, Rev. Geophys. 6, 175-199,
1968.
Barrett, D. R., The genesis of kimberlites
and associated rocks: Strontium isotopic
evidence, in Physics and Chemistry of the
Earth, 9, pp. 637-654, L. H. Ahrens, J. B.
Dawson, A. R. Duncan, and A. J. Erlank,
eds., Elmsford, N. Y., Pergamon Press,
1975.
Bell, K., and J. L. Powell, Strontium
isotopic studies of alkalic rocks: The
potassium-rich lavas of the Birunga and
Toro-Ankole regions, East and Central
Equatorial Africa, J. Petrol., 10, 536-572,
1969.
Bell, K., and J. L. Powell, Strontium
isotopic studies of alkalic rocks: The al-
kalic complexes of eastern Uganda, Geol.
Soc. Amer. Bull., 81, 3481-3490, 1970.
Bence, A.E., The differentiation history of
the Earth by Rb-Sr isotopic relationships,
Ph.D. thesis, Mass. Inst, of Technology,
1966.
Beswick, A. E., K and Rb relations in basalt
and other mantle derived materials: Is
phlogopite the key? Geochim. Cos-
mochim. Acta, 40, in press, 1976.
Biscaye, P. E., and E. J. Dasch, The
rubidium, strontium, strontium-isotope
system in deep-sea sediments: Argentine
basin, J. Geophys. Res., 76, 5087-5096,
1971.
Boettcher, A. L., B. O. Mysen, and P. J.
Modreski, Melting in the mantle: phase
relationships in natural and synthetic
peridotite-rLO and peridotite-H,0-C0 2 at
high pressure, Phys. Chem. Earth, 9, pp.
855-867, L. H. Ahrens, J. B. Dawson, A.
R. Duncan, and A. J. Erlank, eds.,
Elmsford, N.Y., Pergamon Press, 1975.
Boyd, F. R., and P. H. Nixon, Kimberlite
diopsides, Carnegie Inst. Wash. Year Book
68, p. 324, 1970.
Boyd, F. R., and P. H. Nixon, Origins of the
ultramafic nodules from some kimber-
lites of northern Lesotho and the Monas-
tery Mine, South Africa, in Physics and
Chemistry of the Earth, 9, pp. 431-454, L.
H. Ahrens, J. B. Dawson, A. R. Duncan,
and A. J. Erlank, eds., Elmsford, N.Y.,
Pergamon Press, 1975.
Bravo, M. S., and M. J. O'Hara, Partial
melting of phlogopite-bearing synthetic
spinel- and garnet-lherzolites, in Phys.
Chem. Earth, 9, pp. 845-854, L. H.
Ahrens, J. B. Dawson, A. R. Duncan, and
A. J. Erlank, eds., Elmsford, N.Y., Perga-
mon Press, 1975.
Brooks, C, S. R. Hart, and I. Wendt, Realis-
tic use of two-error regression treatments
as applied to rubidium-strontium data,
Rev. Geophys. Space Phys., 10, 551-577,
1972.
Brooks, C, S. R. Hart, A. Hofmann, and D.
E. James, Rb-Sr mantle isochrons from
oceanic regions, Earth Planet. Sci. Lett.,
in press, 1976.
Burwell, A. D. M., Rb-Sr isotope geochemis-
try of lherzolites and their constituent
minerals from Victoria, Australia, Earth
Planet. Sci. Lett., 28, 69-78, 1975.
Chapman, D. S., and H. N. Pollack, Global
heat flow: A new look, Earth Planet. Sci.
Lett., 28 23-32, 1975.
Chow, T. J., and C. Patterson, The occur-
rence and significance of lead isotopes in
pelagic sediments, Geochim. Cosmochim.
Acta, 26, 263-308, 1962.
Church, S. E., Strontium isotope and alkali
element geochemistry of selected sedi-
ments from the northeast Pacific Ocean,
Geochim. Cosmochim. Acta, 35, 1300-
1305, 1971.
Church, S. E., and M. Tatsumoto, Lead
isotope relations in oceanic ridge basalts
from the Juan de Fuca-Gorda ridge area,
northeast Pacific Ocean, Contrib. Min-
eral. Petrol, 53, 253-279, 1975.
DEPARTMENT OF TERRESTRIAL MAGNETISM
203
Church, S. E., and G. R. Tilton, Lead and
strontium isotopic studies in the Cascade
Mountains: Bearing on andesite genesis,
Geol. Soc. Amer. Bull., 84, 431-454, 1973.
Clark, S. P., Jr., and A. E. Ringwood, Den-
sity distribution and constitution of the
mantle, Rev. Geophys. Space Phys., 2,
35-88, 1964.
Compston, W., The Table Hill volcanics of
the Officer Basin, Precambrian or
Paleozoic? Geol. Soc. Australia J., 21,
403-411, 1974.
Compston, W., and L. F. Lovering, The
strontium isotopic geochemistry of
granulitic and eclogitic inclusions from
the basic pipes of Delegate, eastern Aus-
tralia, Geochim. Cosmochim. Acta, 33,
691-699, 1969.
Compston, W., I. MacDougall, and K.
Heier, Geochemical comparison of the
Mesozoic basaltic rocks of Antarctica,
South Africa, South America, and Tas-
mania, Geochim. Cosmochim. Acta, 32,
129-149, 1968.
Cooper, J. A., and D. H. Green, Lead isotope
measurements on lherzolite inclusions
and host basanites from western Victoria,
Australia, Earth Planet. Sci. Lett., 6,
69-76, 1969.
Cormier, R. F., Radiometric dating of the
Coldbrook group of southern New
Brunswick, Canada, Can. J. Earth Sci., 6,
393-398, 1969.
Dasch, E. J., Sr isotopes in weathering pro-
files, deep-sea sediments and sedimen-
tary rocks, Geochim. Cosmochim. Acta,
33, 1521-1552, 1969.
Dasch, E. J., and D. H. Green, Strontium
isotope geochemistry of lherzolite inclu-
sions and host basaltic rocks, Victoria,
Australia, Amer. J. Sci., 275, 461-469,
1975.
Dasch, E. J., C. E. Hedge, and J. Dymond,
Effect of sea water interaction on stron-
tium isotope composition of deep-sea
basalts, Earth Planet. Sci. Lett., 19, 177-
183, 1973.
Davies, R. D., H. L. Allsopp, A. J. Erlank,
and W. I. Manton, Sr-isotopic studies on
various layered intrusions in southern
Africa, Geol. Soc. S. Africa Sp. Pub. 1,
576, 1970.
Dawson, J. B., Kimberlites and their rela-
tionship to the upper mantle, Trans. Roy.
Soc. London, A271, 297-311, 1972.
Doe, B. R., P. W. Lipman, C. E. Hedge, and
H. Kumsawa, Primitive and contami-
nated basalts from the southern Rocky
Mountains, U.S.A., Contrib. Mineral. Pet-
rol. ,21, 142-156, 1969.
Dziewonski, A. M., Upper mantle models
from "pure-path" dispersion data, J.
Geophys. Res., 76, 2587-2601, 1971.
Ewart, A., and J. J. Stipp, Petrogenesis of
the volcanic rocks of the central North Is-
land, New Zealand, as indicated by a
study of H7 Sr/ H,i Sr ratios, and Sr, Rb, K, U,
and Th abundances, Geochim. Cos-
mochim. Acta, 32, 699-736, 1968.
Farhat, J. S., Geochemical investigation of
the early Archean rocks of the Minnesota
River Valley, Ph.D. dissertation, Univ. of
California, Los Angeles, 1975.
Faure, G., R. L. Hill, L. M. Jones, and D. H.
Elliot, Isotope composition of strontium
and silica content of Mesozoic basalt and
dolerite from Antarctica, unpublished
manuscript presented at SCAR Sym-
posium, Oslo, Norway, Aug. 6-15, 1972.
Forbes, W. C, and M. F. J. Flower, Phase
relations of titan-phlogopite, KoMg 2
TiALS^Oi,, (OH) 4 : a refractory phase in
the upper mantle? Earth Planet. Sci.
Lett., 22, 60-66, 1974.
Gast, P. W., Trace element fractionation
and the origin of tholeiitic and alkaline
magma types, Geochim. Cosmochim.
Acta, 32, 1057, 1968.
Gast, P. W., G. R. Tilton, and C. Hedge,
Isotopic composition of lead and Sr from
Ascension and Gough Islands, Science,
145, 1181-1185, 1964.
Girod, M., G. Camus, Y. Vialette, Sur la
presence de tholeiites a Tile Saint-Paul
(Ocean Indien), Contrib. Mineral. Petrol.,
33, 108, 1971.
Goncz, J. H., and J. R. Cleary, Variations in
the structure of the upper mantle beneath
Australia, from Rayleigh observations,
Geophys. J. Roy. Astron. Soc, in press,
1976.
Grant, N. K., J. L. Powell, J. D. Walther,
and F. R. Burkholder, The isotopic com-
position of Sr in lavas from St. Helena,
South Atlantic, in press, 1976.
Green, D. H., Experimental melting studies
on a model upper mantle composition at
high pressure under water-saturated and
water-undersaturated conditions, Earth
Planet. Sci. Lett., 19, 37-53, 1973.
Green, D. H., and A. E. Ringwood, The
genesis of basaltic magmas, Contrib.
Mineral. Petrol., 15, 103-190, 1967.
Green, T. H., and A. E. Ringwood, Genesis
of the calc-alkaline igneous suite, Con-
trib. Mineral. Petrol., 18, 105-162, 1968.
204
CARNEGIE INSTITUTION
Green, T. H., and A. E. Ringwood, High
pressure experimental studies on the ori-
gin of andesites, Oreg. Dept. Geol. Min-
eral. Ind. Bull, 65, 21-32, 1968.
Halpern, M., U. G. Cordani, and M. Be-
renholc, Rev. Brasil. Geociencas, 4, 223,
1974.
Harris, P. G., and E. A. K. Middlemost, The
evolution of kimberlites, Lithos, 3, 77-88,
1969.
Harris, P. G., R. Hutchinson, and D. K.
Paul, Plutonic xenoliths and their rela-
tion to the upper mantle, Phil. Trans.
Roy. Soc. London, A271, 313-323, 1972.
Hart, S. R., K, Rb, Cs content and K/Rb,
K/Cs ratios of fresh and altered sub-
marine basalts, Earth Planet. Sci. Lett., 6,
295-303, 1969.
Hart, S. R., Dredge basalts: Some geochem-
ical aspects (abstr.), Trans. Amer.
Geophys. Union, 52, 376, 1971.
Hart, S. R., Submarine basalts from
Kilauea rift, Hawaii: Nondependence of
trace element composition on extrusion
depth, Earth Planet. Sci. Lett., 20, 201-
203, 1973.
Hart, S. R., LIL-element geochemistry, leg
34 basalts, in Initial Reports of the Deep
Sea Drilling Project, 34, in press, 1976.
Hart, S. R., and C. Brooks, Clinopy-
roxene-matrix partitioning of K, Rb, Cs,
Sr, and Ba, Geochim. Cosmochim. Acta,
38, 1799-1806, 1974.
Hart, S. R., J. G. Schilling, and J. L. Powell,
Basalts from Iceland and along the Reyk-
janes Ridge: Sr isotope geochemistry, Na-
ture Phys. Sci., 246, 104-107, 1973.
Hedge, C. E., Variations in radiogenic
strontium found in volcanic rocks, J.
Geophys. Res., 71, 6119-6126, 1966.
Hedge, C. E., R. A. Hildreth, and W. T.
Henderson, Strontium isotopes in some
Cenozoic lavas from Oregon and Wash-
ington, Earth Planet. Sci. Lett., 8, 434-
438, 1970.
Hedge, C. E., and D. C. Noble, Late
Cenozoic basalts with unusually high
Sr/Rb and K7 Sr/ Kli Sr ratios from the south-
ern Great Basin, western United States,
Geol. Soc. Am. Bull., 82, 3503-3510,
1971.
Hedge, C. E., and Z. E. Peterman, The
strontium isotopic composition of basalts
from the Gorda and Juan de Fuca Rises,
northeastern Pacific Ocean, Contrib.
Mineral. Petrol, 27, 114, 1970.
Hedge, C. E., Z. E. Peterman, and W. R.
Dickinson, Petrogenesis of lavas from
Western Samoa, Geol. Soc. Amer. Bull.
83, 2709-2714, 1972.
Hedge, C. E., N. D. Watkins, R. A. Hildreth,
and W. P. Doering, "Sr/^Sr ratios in
basalts from islands in the Indian Ocean,
Earth Planet. Sci. Lett., 21, 29-34, 1973.
Heier, K. S., W. Compston, and I. McDou-
gall, Thorium and uranium concentra-
tions, and the isotopic composition of
strontium in the differentiated Tasma-
nian dolerites, Geochim. Cosmochim.
Acta, 29, 643-659, 1965.
Hofmann, A. W., Diffusion of Ca and Sr in a
basalt melt, in Carnegie Inst. Wash. Year
Book 74, pp. 183-189, 1975.
Hofmann, A. W., and S. R. Hart, An as-
sessment of local and regional isotopic
equilibrium in a partially molten mantle,
in Carnegie Inst. Wash. Year Book 74, pp.
195-210, 1975.
Hubbard, N. J., Some chemical features of
lava from the Manu'a Islands, Pacific
Sci., 25, 178, 1971.
Hurley, P. M., H. W. Fairbairn, and W. H.
Pinson, Rb-Sr isotopic evidence in the
origin of potash-rich lavas of western
Italy, Earth Planet. Sci. Lett., 5, 301-306,
1966.
Hutchinson, R., and J. B. Dawson, Rb, Sr,
and H7 Sr/ Ki; Sr in ultrabasic xenoliths and
host-rocks, Lashaine Volcano, Tanzania,
Earth Planet. Sci. Lett., 9, 87-92, 1970.
James, D. E., Plate tectonic model for the
evolution of the central Andes, Bull. Geol.
Soc. Am., 82, 3325-3346, 1971.
James, D. E., C. Brooks, and A. Cuyu-
bamba, Andean Cenozoic volcanism:
Magma genesis in the light of strontium
isotopic composition and trace-element
geochemistry, Geol. Soc. Am. Bull, 87,
592-600, 1976.
Jordan, T. H., The continental tectosphere,
Rev. Geophys. Space Phys., 13, 1-12,
1975.
Kanamori, H., Velocity and Q of mantle
waves, Phys. Earth Planet. Inter., 2,
259-275, 1970.
Kay, R., Chemical zonation of the oceanic
mantle (abstr.) Trans. Amer. Geophys.
Union, 56, 1077, 1975.
Kistler, R. W., and Z. E. Peterman, Varia-
tions in Sr, Rb, K, Na, and initial Sr^/Sr™
in Mesozoic granitic rocks and intruded
wall rocks in central California, Geol
Soc. Amer. Bull, 84, 3489-3512, 1973.
Klerkx, J., S. Deutsch, P. DePaepe, Rb, Sr
content and Sr isotopic composition of
strongly alkalic basaltic rocks from the
Cape Verde Islands, Contrib. Mineral.
Petrol, 45, 107-118, 1974.
Kudo, A. M., K. I. Aoki, and D. G. Brookins,
The origin of Pliocene-Holocene basalts of
DEPARTMENT OF TERRESTRIAL MAGNETISM
205
New Mexico in the light of strontium-
isotopic and major element abundances,
Earth Planet. Sci. Lett., 13, 200-204,
1971.
Kudo, A. M., D. G. Brookins, and A. W.
Laughlin, Sr-isotopic disequilibrium in
lherzolites from the Puerco Necks, New
Mexico, Earth Planet. Sci. Lett., 15, 291-
295, 1972.
Laughlin, A. W., D. G. Brookins, and J. R.
Carden, Variations in the initial stron-
tium ratios of a single basalt flow, Earth
Planet. Sci. Lett., 14, 79-82, 1972.
Leeman, W. P., The isotopic composition of
strontium in late-Cenozoic basalts from
the Basin-Range province, western
United States, Geochim. Cosmochim.
Acta, 34, 857-872, 1970.
Leeman, W. P., Late Cenozoic alkali-rich
basalt from the western Grand Canyon
area, Utah and Arizona: Isotopic composi-
tion of strontium, Geol. Soc. Amer. Bull.,
85, 1691-1696, 1974.
Leeman, W. P., Radiogenic tracers applied
to basalt genesis in the Snake River
Plain- Yellowstone National Park re-
gion — evidence for a 2.7 b.y.-old upper
mantle keel, Geol. Soc. Amer. Abstr. 7-1 ,
1165, 1975.
Leeman, W. P., and W. I. Manton, Stron-
tium isotopic composition of basaltic
lavas from the Snake River Plain, south-
ern Idaho, Earth Planet. Sci. Lett., 11,
420-434, 1971.
Lessing, P., and E. J. Catanzaro, Sr H7 /Sr Hi;
in Hawaiian Lavas, J. Geophys. Res., 69,
1599-1601, 1964.
Lovering, J. F., and M. Tatsumoto, Lead
isotopes and the origin of granulite and
eclogite inclusions in deep-seated pipes,
Earth Planet. Sci. Lett., 4, 350-356, 1968.
Lussiaa-Berdou-Polve, M., and P. Vidal, In-
itial Sr isotopic composition of volcanic
rocks from Jan Mayen and Spitsbergen,
Earth Planet. Sci. Lett., 18, 333-338,
1973.
MacGregor, I. D., Mafic and ultramafic in-
clusions as indicators of origin of basaltic
magmas, J. Geophys. Res., 73, 3737-
3745, 1968.
MacGregor, I. D., Petrologic and thermal
structure of the upper mantle beneath
South Africa in the Cretaceous, in
Physics and Chemistry of the Earth, 9, pp.
455-466, L. H. Ahrens, J. B. Dawson, A.
R. Duncan, and A. J. Erlank, eds.,
Elmsford, N.Y., Pergamon Press, 1975.
Manton, W. I., The origin of associated
basic and acid rocks in the Lebombo-
Nuanetsi igneous province, southern Af-
rica, as implied by strontium isotopes, J.
Petrol., 9, 23, 1968.
Manton, W. I., and M. Tatsumoto, Some Pb
and Sr isotopic measurements of eclogites
from the Roberts Victor Mine, South Af-
rica, Earth Planet. Sci. Lett., 10, 217-226,
1971.
Mark, R. K., C. L. Hu, H. R. Bowman, F.
Asaro, E. H. McKee, and R. R. Coats, A
high K7 Sr/ Kll Sr mantle source for low alkali
tholeiite, northern Great Basin, Geochim.
Cosmochim. Acta, 39, 1671-1678, 1975.
McDougall, I., and W. Compston, Sr-isotope
composition and K/Rb ratios in some
rocks from Reunion and Rodriguez, In-
dian Ocean, Nature, 207, 252-253, 1965.
Meijer, A., Pb and Sr isotopic data bearing
on the origin of lavas from the Mariana
Arc System, Geol. Soc. Amer. Bull., in
press, 1976.
Mitchell, R. H., and J. H. Crocket, The
isotopic composition of strontium in some
South African kimberlites, Contrib. Min-
eral. Petrol, 30, 277-290, 1971.
Modreski, P. J., and A. L. Boettcher, Phase
relations of phlogopite in the system
2 -H 2 to 35 kilobars: a better model for
micas in the interior of the earth, Am. J.
Sci., 273, 385, 1973.
Montigny, R., M. Javoy, and C. J. Allegre,
Le probleme des andesites: Etude du vol-
canisme quaternaire du Costa Rica
(Amerique centrale) a l'aide des traceurs
couples 87 Sr/ 8S Sr et 1K 0/ lfi O,Soc. Geol. Fr.
Bull., 11, 794-799, 1969.
Noble, D. C, C. E. Hedge, E. H. McKee, and
M. K. Corringa, Reconnaissance study of
the strontium isotopic composition of
Cenozoic volcanic rocks in the northwest-
ern Great Basin, Geol. Soc. Amer. Bull.,
84, 1393-1406, 1973.
O'Hara, M. J., M. J. Saunders, and E. L. P.
Mercy, Garnet-peridotite, primary ul-
trabasic magma and eclogite: Interpreta-
tion of upper mantle processes in kimber-
lite, in Physics and Chemistry of the
Earth, 9, pp. 571-604, L. H. Ahrens, J. B.
Dawson, A. R. Duncan, and A. J. Erlank,
eds., Elmsford, N. Y., Pergamon Press,
1975.
Okal, E. A., and D. L. Anderson, A study of
lateral inhomogeneities in the upper
mantle by multiple ScS travel-time re-
siduals, Geophys. Res. Lett., 2, 313-316,
1975.
O'Nions, R. K., and R. J. Pankhurst, Pet-
rogenetic significance of isotope and
trace-element variations in volcanic
rocks from the Mid-Atlantic Ridge, J.
Petrol., 15, 603, 1974.
206
CARNEGIE INSTITUTION
Oversby, V. M., and P. W. Gast, Isotopic
composition of lead from oceanic islands,
J. Geophys. Res., 75, 2097-2114, 1970.
Pankhurst, R. J., Strontium isotope studies
related to petrogenesis in the Caledonian
basic igneous province of northeast Scot-
land,,/. Petrol., 10, 115, 1969.
Paul, D. K., Strontium isotope studies on
ultramafic inclusions from Dreiser
Weiher, Eifel, Germany, Contrib. Min-
eral. Petrol, 34, 22-28, 1971.
Peterman, Z. E., I. S. E. Carmichael, and A.
L. Smith, Sr^/Sr*' 1 ratios of Quaternary
lavas of the Cascade Range, northern
California, Geol. Soc. Amer. Bull., 81,
311-318, 1970a.
Peterman, Z. E., I. S. E. Carmichael, and A.
L. Smith, Strontium isotopes in Quater-
nary basalts of southeastern California,
Earth Planet. Sci. Lett., 7, 381-384,
1970b.
Peterman, Z. E., B. R. Doe, and H. H.
Prostka, Lead and strontium isotopes in
rocks of the Absaroka volcanic field,
Wyoming, Contrib. Mineral. Petrol., 27,
121-130, 1970.
Peterman, Z. E., and C. E. Hedge, Related
strontium isotopic and chemical varia-
tions in oceanic basalts, Geol. Soc. Amer.
Bull., 82, 493-499, 1971.
Peterman, Z. E., C. E. Hedge, R. G. Cole-
man, and P. D. Snavely, Jr., Sr^/Sr 811
ratios in some eugeosynclinal sedimen-
tary rocks and their bearing on the origin
of granitic magma in orogenic belts,
Earth Planet. Sci. Lett., 2, 433-439, 1967.
Philpotts, J. A., and C. C. Schnetzler,
Phenocryst-matrix partition coefficients
for K. Rb, Sr., and Ba, with applications
to anorthosite and basalt genesis,
Geochim. Cosmochim. Acta, 34, 307-322,
1970.
Pollack, H. N., and D. S. Chapman, On the
regional variation of heat flow, geo-
therms, and lithospheric thickness, sub-
mitted to Tectonophysics, 1976.
Powell, J. L., and K. Bell, Strontium iso-
topic studies of alkalic rocks. Localities
from Australia, Spain, and the western
United States, Contrib. Mineral. Petrol.,
27, 1-10, 1970.
Powell, J. L., and K. Bell, Isotopic composi-
tion of strontium in alkalic rocks, in The
Alkaline Rocks, p. 412, H. Sorenson, ed.,
New York, Wiley & Sons, 1974.
Powell, J. L., and S. E. Delong, Isotopic
composition of strontium in volcanic
rocks from Oahu, Science, 153, 1239-
1242, 1966.
Pushkar, P., Strontium isotopic ratios in
volcanic rocks of three island arc areas, J.
Geophys. Res., 73, 2701-2714, 1968.
Ramberg, H., Mantle diapirism and its tec-
tonic and magmagenic consequences,
Phys. Earth Planet. Inter., 5, 45-60, 1972.
Rhodes, J. M., and J. B. Dawson, Major and
trace element chemistry of peridotite in-
clusions from the Lashaine Volcano, Tan-
zania, in Physics and Chemistry of the
Earth, 9, pp. 545-558, L. H. Ahrens, J. B.
Dawson, A. R. Duncan, and A. J. Erlank,
eds., Elmsford, N. Y., Pergamon Press,
1975.
Ridley, W. I., and J. B. Dawson, Lithophile
trace element data bearing on the origin
of peridotite xenoliths, ankaramite, and
carbonatite from Lashaine Volcano, N.
Tanzania, inPhysics and Chemistry of the
Earth, 9, pp. 559-570, L. H. Ahrens, J. B.
Dawson, A. R. Duncan, and A. J. Erlank,
eds., Elmsford, N. Y., Pergamon Press,
1975.
Ringwood, A. E., Composition and Petrol-
ogy of the Earth's Mantle, 618 pp., Highs-
town, New Jersey, McGraw-Hill Book
Co., 1975.
Ringwood, A. E., and J. F. Lovering, Sig-
nificance of pyroxene-ilmenite inter-
growths among kimberlite xenoliths,
Earth Planet. Sci. Lett., 7, 371-375, 1970.
Roy, R. F., D. D. Blackwell, and F. Birch,
Heat generation of plutonic rocks and
continental heat-flow provinces, Earth
Planet. Sci. Lett., 5, 1-12, 1968.
Sacks, I. S., Distribution of absorption of
shear waves in South America and its
tectonic significance, in Carnegie Inst.
Wash. Year Book 67, pp. 339-344, 1969.
Sacks, I. S., and H. Okada, A comparison of
the anelasticity structure beneath west-
ern South America and Japan, Phys.
Earth Planet. Inter., 9, 211-219, 1974.
Schilling, J. -G., Sea-floor evolution: rare
earth evidence, Phil. Trans. Roy. Soc.
London, 268, 663, 1971.
Schilling, J. -G., Azores mantle blob: Rare-
earth evidence, Earth Planet. Sci. Lett.,
25, 103-115, 1975.
Sclater, J. G, and J. Francheteau, The im-
plication of terrestrial heat flow observa-
tions on current tectonic and geochemical
models of the crust and upper mantle of
the earth, Geophys. J. Roy. Astron. Soc,
20, 509-542, 1970.
Scott, R. B., R. W. Nesbitt, E. J. Dasch, and
R. L. Armstrong, A strontium isotope evo-
lution model for Cenozoic magma genesis,
DEPARTMENT OF TERRESTRIAL MAGNETISM
207
eastern Great Basin, U.S.A., Bull. Vol-
canol.2-35, 1-26, 1971.
Shimizu, N., An experimental study of the
partitioning of K, Rb, Cs, Sr., and Ba be-
tween clinopyroxene and liquid at high
pressure, Geochim. Cosmochim. Acta, 38,
1789-1798, 1974.
Sipkin, S. A., and T. H. Jordan, Lateral
heterogeneity of the upper mantle deter-
mined from the travel times of ScS, J.
Geophys. Res., 80, 1474-1484, 1975.
Steuber, A. M., and M. Ikramuddin,
Rubidium, strontium, and the isotopic
composition of strontium in ultramafic
nodule minerals and host basalts,
Geochim. Cosmochim. Acta, 38, 207-216,
1974.
Stipp, J. J., The geochronology and pet-
rogenesis of the Cenozoic volcanics of
North Island, New Zealand, unpublished
Ph.D. thesis, Australian National Uni-
versity, Canberra, Australia, 1968.
Subbarao, K. V., The strontium isotopic
composition of basalts from the East
Pacific and Chile Rises and abyssal hills
in the eastern Pacific Ocean, Contrib.
Mineral. Petrol, 37, 111, 1972.
Subbarao, K. V., and C. E. Hedge, K, Rb,
Sr, and K7 Sr/ Ki; Sr in rocks from the Mid-
Indian Ocean Ridge, Earth Planet. Sci.
Lett., 18, 223-228, 1973.
Sun, S. S., and G. N. Hanson, Evolution of
the mantle: Geochemical evidence from
alkali basalt, Geol., 3, 297-302, 1975a.
Sun, S. S., and G. N. Hanson, Origin of Ross
Island basanitoids and limitations upon
the heterogeneity of mantle sources for
alkali basalts and nephelinites, Contrib.
Mineral. Petrol., 52, 77, 1975b.
Sun, S. S., and B. Jahn, Lead and strontium
isotopes in post-glacial basalts from Ice-
land, Nature, 255, 527, 1975.
Tatsumoto, M., Genetic relationships of
ocean basalts as indicated by lead
isotopes, Science, 153, 1094-1101, 1966.
Toksoz, M. N., and D. L. Anderson, Phase
velocities of long-period surface waves
and structure of the upper mantle: 1.
Great-circle Love and Rayleigh wave
data, J. Geophys. Res., 71, 1649-1658,
1966.
Wagner, P. A., The evidence of the kimber-
lite pipes on the constitution of the outer
part of the earth, S. African J. Sci., 25,
127, 1928.
White, W. M., S. R. Hart, and J. G. Schil-
ling, Geochemistry of the Azores and the
Mid-Atlantic Ridge: 29°N to 60°N, in
Carnegie Inst. Wash. Year Book 74, 224-
234, 1975.
Yoder, H. S., Jr., Generation of basaltic
magma, National Academy of Sciences,
in press, 1976.
York, D., Least-squares fitting of a straight
line, Can. J. Phys., 44, 1079-1086, 1966.
Geochemistry of Late Cenozoic
Volcanic Rocks from the Nevado
de Toluca Area, Mexico
D. J. Whitford and K. Bloomfield*
Introduction
Orogenic igneous rocks are charac-
terized by a wide diversity of composi-
tions ranging from island arc tholeiites
through a spectrum of calc-alkaline to
high-K calc-alkaline (or shoshonitic)
varieties. Although they are usually
associated with zones of lithospheric
subduction, they occur in a variety of
tectonic environments, from purely in-
traoceanic island arcs such as the
Tonga-Kermadec island chain to conti-
nental margins such as the Andes of
South America. It is the purpose of this
report to outline the geochemistry of a
suite of orogenic igneous rocks from a
continental margin, to compare these
rocks with those from a typical island
arc, and to outline some of the
geochemical implications and limita-
tions on petrogenetic models.
The major and trace element and Sr
isotopic compositions of a representa-
tive suite of volcanic rocks have been
determined. The rocks, which range in
age from Miocene to Recent, are from
an area of approximately 2500 km 2
centered on the volcano Nevado de To-
luca (Bloomfield and Valastro, 1974).
This volcanic ruin lies within the Mex-
ican volcanic belt approximately 80 km
west^southwest of Mexico City. It is
situated on old, typically continental
crust, approximately 40 km thick
(Cummings and Schiller, 1971). The
volcano, which has been built up on a
series of Tertiary and Mesozoic vol-
* Overseas Division, Institute of Geological
Sciences, London.
208
CARNEGIE INSTITUTION
canics and sediments, consists of an
eroded central strato- volcano composed
mostly of andesitic and dacitic flows
with lahar and fluvial deposits, often
capped by pumice on the lower slopes
(Bloomfield and Valastro, 1974).
Analytical Techniques
Major elements have been deter-
mined by electron microprobe on di-
rectly fused glasses following the
methods of Nicholls (1974a). The trace
elements Rb, Sr, Zr, and Y were deter-
mined by x-ray fluorescence spec-
trometry on pressed powders following
the methods of Norrish and Chappell
(1967). Major and trace element
analyses were performed at the Re-
search School of Earth Sciences, Aus-
tralian National University.
The Sr isotopic composition of the
lavas was determined at DTM on a
6-inch automated mass spectrometer.
All 87 Sr/ 86 Sr ratios have been nor-
malized to an 86 Sr/ 88 Sr ratio of 0.1194,
and all analyses have been corrected to
an E and A SrC0 3 standard 87 Sr/ 86 Sr
ratio of 0.70800. Repeated analysis of
the E and A standard during the time
in which the analyses were performed
yielded an average 87 Sr/ 86 Sr ratio of
0.707934 ± 0.000044 (2a of popula-
tion.
Major and Trace Elements
Results. Selected representative
analyses are listed in Table 11. The
lavas exhibit many features character-
istic of orogenic igneous rocks, includ-
ing generally low concentrations of
Ti0 2 and MgO and high concentrations
of A1 2 3 . Concentrations of K 2 0, Rb,
and Zr and the FeO-MgO relationships
indicate calc-alkaline affinities (Jakes
and Gill, 1970), which are typical of the
volcanic rocks of continental margins.
However, in many important re-
spects they differ from typical island
arc volcanics such as those found in the
western Pacific. The Mexican rocks are
distinctly richer in Si0 2 , Na 2 0, and
MgO relative to FeO. In these respects,
they are similar to some Recent vol-
canics from the Andes (Noble et al.,
1975). Some of these differences are
summarized in Figs. 34 and 35
Figure 34 illustrates the frequency
distribution of Si0 2 concentrations in
both the Mexican lavas and a represen-
tative suite of typical island arc lavas
from Indonesia (Nicholls and Whitford,
1976). The Mexican rocks have Si0 2
concentrations ranging from 55% to
70% with a mode of about 67%. Basalts
(< 52% Si0 2 ) are totally absent. This
contrasts with the Indonesian rocks,
which have a mode of approximately
TABLE 11. Selected Representative Major and Trace Element Analyses
157-72
133A-72
91-72
183-72
134-72
161-72
9-72
'SiO,
55.1
57.6
59.4
62.4
64.1
66.2
67.0
TiO,
0.74
0.96
0.84
0.90
0.68
0.57
0.50
Al,6,
16.4
16.2
17.1
16.7
18.3
16.9
16.8
2 FeO
7.43
6.49
6.08
5.08
4.21
3.62
3.41
MO
<0.12
<0.12
<0.12
<0.12
<0.12
<0.12
<0.12
MgO
7.94
6.86
4.86
4.04
2.20
1.97
1.96
CaO
8.18
6.66
6.05
5.11
4.44
4.10
3.97
Na,0
3.31
3.62
4.11
3.71
4.55
4.57
4.47
K,6
0.82
1.68
1.55
2.07
1.50
2.14
1.98
Mg/Mg + Fe* +
0.66
0.65
0.59
0.59
0.48
0.49
0.50
'Rb
15.7
34.6
28.4
48.2
14.2
51.3
42.9
Sr
366
541
537
448
518
570
516
Zr
100
182
142
244
158
158
161
Y
59
23
21
25
17
16
15
X A11 values in wt % normalized to 100% total.
2 Total Fe expressed as FeO.
3 A11 trace element concentrations expressed as ppm.
DEPARTMENT OF TERRESTRIAL MAGNETISM
209
Nevado de Toluco
Sunda arc- Indonesia
S1O2 (wt %)
Fig. 34. Histograms of Si0 2 (wt % ) contents of
analyzed lavas from the Nevado de Toluca area
and from the Sunda arc of Indonesia (Nicholls
and Whitford, 1976).
vado de Toluca
j Sunda arc -Indonesia
Mcj/(Mg+Fe Hmol )
Fig. 35. Histograms of the atomic ratio
Mg/Mg + Fe 2+ in analyzed lavas from the
Nevado de Toluca area and from the Sunda arc of
Indonesia (Nicholls and Whitford, 1976). Before
calculation of Mg/Mg + Fe 2 " 1 ", ferrous iron in all
analyses was adjusted to FeO = 0.85 £ FeO (total
Fe as FeO). This relationship is similar to that
determined for basaltic to andesitic liquids
equilibrated at oxygen fugacities of the
wiistite-magnetite buffer (Fudali, 1965); similar
oxygen fugacities are believed to prevail in the
upper mantle (Boettcher, 1973).
55% Si0 2 . In this respect the Mexican
volcanics are very similar to the
Cenozoic lavas from Peru (D. E. James,
personal communication).
Figure 35 shows the frequency dis-
tribution of the atomic ratio Mg/Mg +
Fe 2 "^ in both the Mexican and Indone-
sian rocks. Although the division is not
as well defined as in Fig. 34, the Mexi-
can rocks are, on average, more mafic
in terms of the ratio Mg/Mg + Fe 2+
than those from Indonesia. Experimen-
tal data for the partitioning of Fe 2+ and
Mg between olivine and coexisting sili-
cate liquid (Green and Ringwood, 1967;
Roeder and Emslie, 1970; Nicholls,
19746) indicate that liquids in equilib-
rium with olivine of supposed upper
mantle peridotite must be highly
magnesian (Mg/Mg + Fe 2+ ^ 0.61)
(Nicholls and Whitford, 1976). Depend-
ing upon both total pressure and water
content, experimental evidence indi-
cates that these equilibrium liquids
may have up to 60% Si0 2 (Nicholls,
19746). These results suggest that de-
spite being relatively silicic, rocks such
as 157-72 and 133A-72 (Table 11) may
represent unfractionated partial melt-
ing products of the mantle.
The magnesian character of the Mex-
ican volcanics together with the Y
abundances, which suggest heavy rare
earth concentrations 7-30 times chon-
dritic abundances, also argues against
any single-stage petrogenetic model
involving partial melting of quartz ec-
logite (Nicholls and Whitford, 1976,
and references therein).
Sr Isotopes
Table 12 lists the concentrations of
SiO,, K,0, Rb, and Sr together with the
ratios K/Rb, Rb/Sr, and 87 Sr/ 86 Sr. Sr
exhibits a significant spread in isotopic
composition, with 87 Sr/ 86 Sr ratios rang-
ing from 0.70321 to 0.70446. Because of
uncertainties about the absolute ages
of the older samples, no corrections
have been applied for the 87 Sr that has
accumulated since eruption. Assuming
an extreme age of 25 m.y. for the oldest
samples, the Rb/Sr ratios are suffi-
ciently low that the differences be-
tween the measured and calculated ini-
tial ratios lie within the uncertainties
of the isotopic analyses.
It is important to note that rocks
which are related in both space and
time have significantly different
isotopic compositions. The 87 Sr/ 86 Sr
ratios are plotted as histograms accord-
ing to age in Fig. 36. On average, there
is a poorly defined overall decrease in
87g r /86g r ra ti with increasing age.
The observed range of isotopic com-
positions is similar to that observed in
volcanic rocks from many other island
210
CARNEGIE INSTITUTION
TABLE 12. Trace Element and Sr Isotopic Results
Age Sample
Si0 2 (%)
K 2 0(%)
Rb
Sr
K/Rb
Rb/Sr
87 Sr/ 86 Sr *
±2cr
Miocene 91-72
59.4
1.55
28.4
537
453
0.0529
0.70374
0.00009
221-72
61.0
1.57
30.7
667
425
0.0460
0.70385
0.00010
172-72
63.5
1.79
20.4
500
728
0.0408
0.70420
0.00009
125-72
64.0
1.90
38.1
621
414
0.0614
0.70368
0.00008
t 107-72
64.3
2.02
39.6
609
423
0.0650
0.70372
0.00011
1 116-72
64.6
1.57
25.7
754
507
0.0341
0.70321
0.00005
45-72
65.4
1.87
46.6
432
333
0.108
147-72
69.0
2.50
50.5
542
411
0.0930
Pliocene 208B-72
64.6
1.70
26.7
478
529
0.0559
0.70376
0.00005
162-72
65.4
2.20
163
668
112
0.244
.70385
0.00007
161-72
66.2
2.14
51.3
570
346
0.0900
0.70401
0.00005
47-72
67.0
2.24
57.8
445
322
0.130
.70391
0.00006
Pleistocene 157-72
55.1
0.82
15.7
366
434
0.0429
0.70381
0.00009
186-72
55.1
1.15
20.5
657
466
0.0312
0.70378
0.00005
132A-72
57.0
2.37
32.8
1330
600
0.0247
0.70419
0.00005
133A-72
57.6
1.68
34.6
541
403
0.0640
181-72
60.8
1.67
36.3
540
382
0.0672
0.70420
0.00007
236-72
62.0
1.76
36.1
593
405
0.0609
0.70421
0.00007
183-72
62.4
2.07
48.2
448
357
0.108
134-72
64.1
1.50
14.2
518
877
0.0274
0.70446
0.00005
1 10-72
65.3
1.97
39.0
681
419
0.0573
0.70379
0.00008
44-72
66.1
1.99
41.2
525
401
0.0785
153-72
66.7
1.92
40.7
516
392
0.0789
0.70409
0.00009
196-72
66.9
2.04
42.2
503
401
0.0839
0.70400
0.00005
Recent 89-72
60.9
1.71
30.3
644
469
0.0470
tVIII/A
63.1
1.58
30.1
528
436
0.0570
0.70358
0.00005
tVIII/5
66.1
1.89
41.9
512
374
0.0818
0.70400
0.00012
t27-71
66.1
1.92
42.2
528
378
0.0799
0.70412
0.00011
1108-72
66.6
2.05
42.3
550
402
0.0769
0.70397
0.00004
t9-72
67.0
1.98
42.9
516
383
0.0831
0.70408
0.00005
t Samples used in Fig. 39.
Corrected to £ and A 87 Sr/ 86 Sr = 0.70800.
i r
n
Recent —
H
Miocene -
_l I L
0.7035 7040 7045
87 Sr/ 86 Sr
Fig. 36. Histograms showing the frequency distribution of K7 Sr/*"Sr ratios in the volcanic rocks
from the Nevado de Toluca area. Average values are indicated by arrows.
DEPARTMENT OF TERRESTRIAL MAGNETISM
211
arcs and continental margins including
the Cascades, most of the northwestern
Pacific arcs, and Japan (Whitford,
1975, and references therein). Despite
some similarities in major and trace
element composition, the Mexican
rocks have 87 Sr/ 86 Sr ratios generally
lower than those from the Andes
(James e* al., 1974a, 19746).
Figure 37 is a plot of 87 Sr/ 86 Sr versus
87 Rb/ 86 Sr in the analyzed lavas from
Nevado de Toluca. Overall, there is
clearly no correlation. However, if only
those samples from the immediate vic-
inity of the volcano itself are chosen,
there is a significant positive correla-
tion (Fig. 38). Using the method of
York (1966), an "age" of 430 ±97 m.y. is
indicated (A 87 Rb = 1.39 x 10" n yr" 1 ).
It may be significant that the age of
approximately 400 m.y. is similar to
that observed by James et al. (19746)
for the Arequipa and Barroso volcanics
of Peru. The fact that the rocks from
Nevado de Toluca have been erupted
within an area of approximately 100
km 2 over a time interval of up to 25
m.y. obviously has important implica-
tions in terms of magma genesis and
eruption. This theme has been de-
veloped in some detail in the report by
Brooks et al. (this Year Book). Apart
from this relationship, the Mexican
lavas show no correlation between
87g r /86g r anc i e ither major or trace ele-
ments.
Discussion
Without knowledge of the nature of
the deep crust beneath Nevado de To-
luca, quantitative evaluation of the
possible importance of crustal melting
and/or assimilation is not possible.
However, it is clear that the relatively
low 87 Sr/ 86 Sr ratios in the volcanics and
their high Sr concentrations place very
severe limitations on such models and
that the ultimate origin of these rocks
lies in the mantle. Despite higher 87 Sr/
86 Sr ratios in many of the Recent vol-
canics from Peru, James et al. (19746)
were forced to the similar conclusion
that crustal processes were not primar-
ily responsible for the observed
geochemical variations.
This immediately raises the question
of why volcanic rocks from at least
some continental margins differ from
those of island arcs. Assuming that the
process of subduction and the sub-
ducted lithosphere itself are similar in
each environment, the differences in
the erupted rocks must therefore re-
0.7045
0.7025
0.1 0.2 0.3 0.4 0.5 0.6 0.7
87 Rb/ 86 Sr
Fig. 37. Plot of K7 Sr/ KI, Sr vs. 87 Kb/ 86 Sr in all analyzed lavas from the Nevado de Toluca area.
212
CARNEGIE INSTITUTION
0. 7040
0.7035
0.7030
0.7025
430±97my
0.70265 ±0.00012
1
0.
0.2
0.3
S7 Rb/ 86 Sr
Fig. 38. Plot of K7 Sr/ Kli Sr vs. "Rb/^Sr for lavas from the immediate vicinity of the volcano Nevado
de Toluca. Two u error limits in the isotopic determinations are also plotted.
fleet differences in the chemistry
and/or the physical state of the mantle
wedge above the subducted litho-
sphere. Sacks and Okada (1974)
showed that the mantle wedge above
the subducted lithosphere beneath the
Andes is itself lithospheric rather than
asthenospheric. The possibility of
lithospheric thickening beneath older
continental areas has been invoked by
Brooks et al. (this Year Book) as an ex-
planation for certain Sr isotopic effects
such as the existence of initial iso-
chrons similar to that shown in Fig. 38.
Although presently unconstrained,
lithospheric thickening beneath conti-
nents offers a solution to the problem of
generating differences between the
mantle source regions beneath island
arcs and some continental margins
Conclusions
1. Late Cenozoic calc-alkaline vol-
canic rocks from the Nevado de Toluca
area of Mexico display features such as
low concentrations of TiQ> which are
characteristic of orogenic igneous
rocks. However, they differ from typi-
cal island arc rocks in having generally
higher concentrations of Si0 2 and
Na,0 and higher Mg/Mg + Fe 2+ ratios.
2. The Mexican volcanics have rela-
tively low 87 Sr/ 86 Sr ratios, ranging
from 0.7032 to 0.7045. There are signif-
icant differences in 87 Sr/ 86 Sr ratios be-
tween spatially and temporally related
rocks. Miocene to Recent rocks from
the Nevado de Toluca volcano itself
have an apparent age of 425 m.y.
3. The low 87 Sr/ 86 Sr ratios and the
higher Mg/Mg + Fe 2+ ratios in at least
some of the volcanics suggest an ulti-
mate origin within the mantle. Dif-
ferences between continental margin
and island arc volcanic rocks may re-
flect lithospheric thickening beneath
older continental regions.
References
Bloomfield, K., and S. Valastro, Late Pleis-
tocene eruptive history of Nevado de To-
DEPARTMENT OF TERRESTRIAL MAGNETISM
213
luca volcano, Central Mexico, Geol. Soc.
Am. Bull., 85, 901-906, 1974.
Boettcher, A. L., Volcanism and orogenic
belts — the origin of andesites, Tectono-
physics, 17, 223-240, 1973.
Cummings, D., and G. I. Schiller, Isopach
map of the earth's crust, Earth Sci. Rev.,
7, 97-125, 1971.
Fudali, R. F., Oxygen fugacities of basaltic
and andesitic magmas, Geochim. Cos-
mochim. Acta, 29, 1063-1075, 1965.
Green, D. H., and A. E. Ringwood, The
genesis of basaltic magmas, Contrib.
Mineral. Petrol., 15, 103-190, 1967.
Jakes, P., and J. Gill, Rare earth elements
and the island arc tholeiitic series, Earth
Planet, Sci. Lett., 9, 17-28, 1970.
James, D. E., C. Brooks, and A. Cuyu-
bamba, Strontium isotopic composition
and K, Rb, Sr geochemistry of Mesozoic
volcanic rocks of the Central Andes, in
Carnegie Inst. Wash. Year Book 73, pp.
970-983, 1974a.
James, D. E., C. Brooks, and A. Cuyu-
bamba, Andean Cenozoic volcanism:
magma genesis in the light of Sr isotopic
composition and trace element
geochemistry, in Carnegie Inst. Wash.
Year Book 73, pp. 983-997, 19746.
Nicholls, I. A., A direct fusion method of
preparing silicate rock glasses for energy
dispersive electron microprobe analysis,
Chem. Geol, 14, 151-157, 1974a.
Nicholls, I. A., Liquids in equilibrium with
peridotitic mineral assemblages at high
water pressures, Contrib. Mineral. Petrol.
45, 289-316, 19746.
Nicholls, I. A., and D. J. Whitford, Primary
magmas associated with Quaternary vol-
canism in the western Sunda arc, In-
donesia, in Volcanism in Australasia, pp.
77-90, R. W. Johnson, ed., N. Y., Elsevier
Publ. Co., 1976.
Noble, D. C, H. R. Bowman, A. J. Hebert,
M. L. Silberman, C. E. Heropoulos, B. P.
Fabbi, and C. E. Hedge, Chemical and
isotopic constraints on the origin of low-
silica latite and andesite from the Andes
of central Peru, Geology, 3, 501-504,
1975.
Norrish, K., and B. W. Chappell, X-ray
fluorescence spectrography, in Physical
Methods in Determinative Mineralogy, pp.
161-214, J. Zussman, ed., N." Y.,
Academic Press, 1967.
Roeder, P. L., and R. F. Emslie, Olivine liq-
uid equilibrium, Contrib. Mineral. Pet-
rol, 29, 275-289, 1970.
Sacks, I. S. and H. Okada, A comparison of
the anelasticity structure beneath west-
ern South America and Japan, Phys.
Earth Planet, Inter., 9, 211-219, 1974.
Whitford, D. J., Strontium isotopic studies
of the volcanic rocks of the Sunda arc, In-
donesia, and their petrogenetic implica-
tions, Geochim. Cosmochim. Acta, 39,
1287-1302, 1975.
York, D., Least squares fitting of a straight
line, Can. J. Phys., 44, 1079-1086, 1966.
Preliminary Rb/Sr Data on the
Minimum Age of the Central
Andean Precambrian Basement
Complex
D. E. James and C. Brooks
Introduction
The Andean volcanic arc obscures
both figuratively and literally the Pre-
cambrian basement of that part of
South America. It is generally
accepted, however, that the orogenic
belt in Peru is underlain at least in
part by the westward extension of the
Brazilian shield (e.g., Stewart, et al.,
1974; Isaacson, 1975), although this
has not been established for adjacent
northern Chile (Gonzales-Bonorino
and Aguirre, 1970). The basement
complex (see Fig. 39), consisting
primarily of granite gneisses and mig-
matites, is best exposed to the west of
and on the westernmost flank of the
Cenozoic volcanic belt (Cobbing and
Pitcher, 1972; Isaacson, 1975). There
imple evidence to suggest that much of
the central Andean volcanic arc is
underlain by ancient sialic crustal
remnants, although the gneiss massif
cannot constitute the whole of the
lower crust as Cobbing and Pitcher
speculate — if for no other reason than
that the crust was probably less than
half its present thickness 100 m.y. ago
(James, 1971).
In studies reported in earlier Year
Books we have concerned ourselves
with the Peruvian coastal gneiss com-
plex only enough to show that it cannot
have contributed in any significant
way to the contamination of the
214
76°
12° -
1 4°
16'
18° -
74 c
72 c
CARNEGIE INSTITUTION
70° 68°
Lima
Precambnan Basement (schematic)
I2 C
I4 C
16°
20 c
78°
76°
74°
72°
70°
Fig. 39. Map giving schematic outline of region in which the coastal Precambrian gneiss complex
crops out in southern Peru. Sample localities are shown for the two samples analyzed in this study.
Mesozoic and Cenozoic volcanic rocks.
This neglect, however, is not to deny
the considerable interest we have in
knowing the age and composition of the
basement complex. Its age is of particu-
lar importance because it bears upon
the question of the age of the litho-
spheric mantle that underlies that part
of the South American continent. We
have proposed (James et al., 1976) that
the anomalous isotopic character of the
late Cenozoic magmas of the central
Andes is inherited from ancient sub-
continental lithosphere with a compo-
sition significantly different from that
of other parts of the mantle. The only
practical means available for placing
age constraints on mantle rocks of the
continental lithosphere is to determine
the history of the oldest rocks exposed
at the surface.
The Rb-Sr measurements given
below are ultimately to be used in con-
junction with zircon age measurements
not yet completed. We consider the
Rb-Sr ages to be minimum or
metamorphic ages. Ages were obtained
by cutting slabs about 1 cm thick and
15 cm across from the samples perpen-
dicular to the banding and measuring
Rb and Sr concentrations (by x-ray
fluorescence) and 87 Sr/ 8,; Sr ratios in
each individual band.
Results
The results of the analyses are plot-
ted in Fig. 40. Sample PE 210 is from
the Charcani gneiss exposed in the Rio
Chili River valley near Arequipa,
Peru, which represents one of the most
easterly exposures of the Peru coastal
gneiss complex. Its Rb-Sr age of
1012 ±52 m.y. is significantly greater
than the K/Ar ages of about 650 m.y.
obtained by Stewart et al. (1974) for
two gneiss samples from the same gen-
eral region of Arequipa. It is remark-
able that PE 210 still retains a 1 b.y.
age despite the fact that the outcrop
from which it was collected is tightly
sandwiched between opposing flanks of
Misti and Chachani, two of the largest
DEPARTMENT OF TERRESTRIAL MAGNETISM
215
0.780r
0.770
0.760-
0.750-
«o° 0.740
<x> 0.730
0.720
0.710
1 1 1 1
1 1 1
1 1 1
-
^^-"o
*^
o PE21CXCharcani gneiss)
• PE111
-
0.7195
1 1 1 1
1 1 1
1 1 1 1
0.10
0.30
0.50
0.70
0.90
1. 10
Rb/Sr
Fig. 40. Plot of K7 Sr/ Nli Sr vs. Rb/Sr for gneiss slab sections. Age and errors calculated using York
regression. Solid dots are for PE 111, open circles for PE 210.
andesitic strato-volcanoes of southern
Peru. This Rb-Sr age further pushes
back the minimum age of the central
Andean basement complex.
Sample PE 111 is from a coastal out-
crop south of Pisco on the Paracas
Peninsula. The outcrop is cut by veins
and dikes of basaltic material, indicat-
ing that the gneiss was intimately in-
volved with Mesozoic or Cenozoic
magmatism. The scatter in the Rb-Sr
data may be due in part to the effects of
this later magmatism. We calculate an
age of 671 ±138 m.y. from the best-fit
line through the data but place little
reliance on this result. The basis of our
skepticism is found in the considerable
data scatter and in the progressive
change of initial K7 Sr/ K,i Sr with position
in the slab (Fig. 41). The points on the
isochron diagram that tend to lie below
the best-fit line come from one end of
the slab; points that lie above the line
come from the other end of the slab,
irrespective of Rb/Sr ratio. This corre-
lation could, for example, be caused by
mixing of the gneiss material with in-
jected material that is isotopically dis-
tinct. Such an interpretation is
suggested by prominent veining found
in the outcrop and in the sample itself.
Whatever the explanation, the slab is
not in isotopic equilibrium for a given
age from one end to the other. Because
of this, the Rb-Sr age measurements on
PE 111 are necessarily suspect, and
hence the difference in isotopic age be-
tween PE 111 and PE 210 is not neces-
sarily real.
References
Cobbing, E. J., and W. S. Pitcher, Plate tec-
tonics and the Peruvian Andes, Nature,
240, No. 99, 51-53, 1972.
Gonzales-Bonorino, F., and L. Aguirre,
Metamorphic facies series of the crystal-
line basement of Chile, Geol.
Rundschau, 59, 979-994, 1970.
Isaacson, P. E., Evidence for a western ex-
tracontinental land source during the
Devonian period in the central Andes,
Geol. Soc. Am. Bull., 86, 39-46, 1975.
James, D. E., Plate tectonic model for the
evolution of the central Andes, Geol. Soc.
Am. Bull., 82, 3325-3346, 1971.
216
CARNEGIE INSTITUTION
0.7280
O
tr>
CD
^ 0.7270
CO
0.7260
1 1 1
E ^
o c
1
I / o"
o -"^ ».
H
°A
Slab
PEIII f\ key
-
1 1 1
1
5 10
Position normal to banding (in cm across slab)
15
Fig. 41. Plot of initial "Sr/^Sr vs. position in the slab. Slab cross section is shown in inset. Letters
key layers analyzed to the K7 Sr/ K,; Sr initial ratios. Abscissa is in centimeters from center of layer A
along a line normal to the layering (see inset).
James, D. E., C. Brooks, and A.
Cuyubamba, Andean Cenozoic vol-
canism: Magma genesis in the light of
strontium isotopic composition and
trace-element geochemistry, Geol. Soc.
Am. Bull., 87, 592-600, 1976.
Stewart, J. W., J. F. Evernden, and N. J.
Snelling, Age determinations from An-
dean Peru: A reconnaissance survey,
Geol. Soc. Am. Bull., 85, 1107-1116,
1974.
Volcanoes, Q, and Seismicity in
Western South America
I. S. Sacks, J. A. Snoke, and A. T. Linde
Introduction
Subduction zones are regions where
oceanic plates, generated at mid-ocean
ridges, plunge into the asthenosphere
under an adjacent oceanic or continen-
tal plate. This interaction generates a
characteristic pattern of seismic and
volcanic activity as well as a clearly
recognizable anelasticity structure. We
have studied the distribution of these
features along the extensive subduc-
tion zone of western South America,
using their expression in the Japan re-
gion as a reference for normal distribu-
tion. Regions of anomalous anelasticity
structure are also found to have diffuse
seismicity and an absence of volcanoes.
The study of the anelasticity struc-
ture of various parts of the earth has
been our major focus for a number of
years (see Year Book 72, pp. 226-233;
Year Book 74, pp. 256-266). In particu-
lar, from broadband seismographs in-
stalled in South America and in Japan,
we have learned that there is a thick,
low absorption (high-Q) keel beneath
the South American continent. This
has been confirmed by a technique
using converted shear to compressional
energy, which is described in the fol-
lowing paper. From the study in the
Japan region (Sacks and Okada, 1974),
we learned, inter alia, that the subdue-
DEPARTMENT OF TERRESTRIAL MAGNETISM
217
tion of a cool oceanic lithosphere can
reduce the amount of partial melt in
the adjacent asthenosphere. This
causes relatively rapid thickening of
the continental lithosphere and re-
duces the thermal gradient in the
downgoing lithosphere. A possible
manifestation of these effects in the
subduction zone of western South
America is described in this report.
Anelasticity . The absorptive part of
seismic energy propagation may be
used to calculate an anelasticity struc-
ture of a region. The effect of an ab-
sorptive medium on seismic waves is to
attenuate the amplitudes at higher
frequencies more than at lower. Seis-
mographs with a wide frequency range
(Sacks, 1966) are used and allow accu-
rate absorption (1/Q) determinations to
be made. Records from local seismo-
graphs, though of smaller bandwidth,
are used to expand the coverage. Spec-
tral ratio techniques (e.g., Sacks, 1968)
are used throughout to avoid depend-
ence on the source spectra of the
earthquakes used for the study.
Anelasticity is often used as a sensitive
indicator of the lithosphere (high-Q, or
low absorption) and the asthenosphere
(low-Q, or high absorption).
Anelasticity studies of the upper
mantle have been made for regions be-
neath Japan (Katsumata, 1960; Utsu,
1966; Sacks and Okada, 1974), Fiji
(Isacks et al., 1968; Barazangi and
Isacks, 1971), South America (Sacks,
1969, Sacks and Okada, 1974), and
New Zealand (Mooney, 1970). The
anelasticity structure of a typical sub-
duction zone consists of a high-Q (Q p ~
1000) lithosphere overlying a low-Q
(Q p — 200-500) asthenosphere, which in
turn overlies the dipping subducted
oceanic lithosphere with Q tJ of 1000
near the surface and 3000 or more at
depth. The lithosphere beneath Japan
is about 70 km thick, whereas that be-
neath South America is in excess of 300
km. The study leading to these conclu-
sions is fully described in Sacks and
Okada (1974), and the gross anelastic-
ity structures are shown in Fig. 42.
MAT
200 400 600
T7~
»2i S 2C6V1 1 OOQ3a*U3T31I
21V'-537S?f—
cuz
1400 KM
r-J O -4"
200 400V^ 600 800 - 10OO 1200 T400 >
'°°<V&;. £ ;■ ,
1000
Fig. 42. The Q,,-structure beneath Japan (left) and South America (right). Dots indicate areas of
high Q (1000-3000); slant lines, intermediate Q- values (300-500); and cross-hatching, low Q- values
(50-100). The small numbers in the dotted region indicate earthquake hypocenters. The region above
the dipping seismic plane is of intermediate Q in Japan, whereas in South America it has high Q. The
high-Q slab in Japan is continuous down to the deepest earthquakes; in South America there is a
substantial decrease in Q below 350 km even though there are earthquakes 600 km deep. In northern
Chile and southern Peru, a distinct low-Q region, marked C, overlies the dipping plane.
218
CARNEGIE INSTITUTION
In some regions, such as northern
Chile and southern Peru, the gross
anelasticity structure is similar to that
beneath Japan except for the much
thicker lithosphere beneath South
America. In these regions, a low-Q zone
was found between the high-Q subduc-
ting lithosphere and the high-Q conti-
nental lithosphere. Intermediate depth
(200-300 km) earthquakes occurring in
the subducting slab were found to have
low-Q paths to the broadband seismo-
graph at Toconce, which is 210 km
from the coast in northern Chile. How-
ever, paths to the same seismograph
originating at, or penetrating to, the
same depth but staying wholly within
the continental lithosphere have high
Q. Fig. 43 shows these seismic ray
paths. Paths to a coastal seismograph
at Antofagasta have high Q from the
same intermediate depth subduction
zone earthquakes which had low Q at
Toconce. Therefore, there must be a
low-Q region between the subducting
lithosphere and the continental litho-
sphere. It is difficult to determine the
thickness of this low-Q zone; it is less
than 100 km but probably not less than
40 km. This zone has a Q„ of about 400,
which is in the range of asthenosphere
values found elsewhere, e.g., Japan.
The low-Q zone can be traced from
200 400 600
Fig. 43. A seismicity section 50 km wide,
normal to the trench in northern Chile. The
numbers indicate earthquake locations (accord-
ing to CGS) and the arrows show the position of
two seismographs whose records were compared.
TCC (Toconce) arrivals from 200-250 km events
show asthenosphere-type Q- values.
Puno in southern Peru to Peldehue in
central Chile. Seismograms from these
stations show the same absorption dis-
tribution as those from Toconce in
northern Chile.
Beneath central Peru, which has
similar seismic station coverage, we
have not been able to find an equiva-
lent low-Q zone. All paths from
earthquakes at depths less than 300
km to the broadband seismograph at
Cusco (340 km from the coast) have
high Q. This seems to be true also for
the Huancayo seismograph farther to
the northwest. These paths are consid-
ered in greater detail in Sacks (1969).
The anelasticity structure described
above is in agreement with the areal
distribution of absorption given by
Barazangi et al. (1975).
In summary, south of Cusco (i.e.,
southern Peru and northern Chile)
there is a low-Q zone between the sub-
ducting lithosphere and the continen-
tal lithosphere; Q values in this zone
are similar to those found in the as-
thenosphere. In the region north of
Cusco, no low-Q zone between the
high-Q subducting and continental
lithospheres has been found, presuma-
bly because such a zone is either very
thin or nonexistent.
Seismicity. The seismicity of subduc-
tion regions is characterized by a dip-
ping seismic plane some tens of
kilometers thick, which is thought to
delineate the upper surface of the sub-
ducting lithospheric plate. There are
no earthquakes in the asthenosphere
on either side of this subducting plate.
Earthquakes do occur in the continen-
tal lithosphere above the subduction.
The seismicity is not uniform along the
subducting plate; regions of low or zero
seismic release are found. In some re-
gions, such as New Hebrides, the
deeper part of the subducted slab ap-
pears to be detached (Barazangi et al.,
1973). This is suggested to be the case
beneath South America as well (Sacks
and Okada, 1974; Snoke et al, 1974,
and others). However, Isacks and
Barazangi (1973) interpret the sub-
ducting plate to be continuous, though
DEPARTMENT OF TERRESTRIAL MAGNETISM
219
having zero seismicity, in the depth
range 350-500 km.
For regions in which the subduction
is not clearly delineated by the seismic-
ity, an alternative method developed
by Okada (1974) may be used to detect
the upper surface of the descending
plate. This technique relies on the effi-
cient conversion of near vertical shear
waves, ScS, into compressional waves
at an inclined interface. Figure 44
shows the ray paths giving rise to the
phase ScSp. The time difference be-
tween ScSp and ScS gives a locus of
possible conversion points in the earth,
indicated by the shaded line in Fig. 45,
which also shows an example of the
phase. The amplitude ratio ScSp/ScS is
used to determine the depth, indicated
by a circle, at which the conversion of
shear-to-compressional energy takes
place (Snoke and Sacks, 1975; Okada,
1976). This method has been confirmed
in areas where the subducting plate is
clearly delineated by earthquakes, e.g.,
Japan and southern Peru (Okada,
1974).
At Peldehue in central Chile, clear
ScSp arrivals originate from the
deeper part (150 km) of the dipping
seismic plane. The interpretation of the
Arequipa (southern Peru) records indi-
cates an interface that coincides well
with that delineated by seismicity.
Farther north, however, this is no
longer the case. The Nana (central
Peru) seismograms show clear ScSp
arrivals (Fig. 45), which also indicate
the probable conversion point. This
conversion point lies on an extension of
the shallow seismicity (near the
trench) which seems to disappear
below 100-km depth. This suggests
that the subduction has a dip angle
similar to the other regions discussed
(southern Peru to central Chile) but is
of low seismicity.
There is considerable seismic activ-
ity in what we have interpreted to be
the continental lithosphere. This is in-
terpreted by some, e.g., Stauder (1975)
and Isacks and Molnar (1971), to be
subducting oceanic lithosphere with a
dip of about 15°. The ScSp observations
and the trend of shallow seismicity
persuade us that this is not the case.
Volcanism. Volcanoes generally
occur in subduction regions at some
SOu Th
4M£ R/(
Station
• Hypocenter
*♦ Earthquakes
CORE
Fig. 44. A schematic model showing the paths of the ScSp phase (solid line) and the ScS phase
(dotted line). The shear energy reflected from the earth's core, ScS, is efficiently converted to com-
pressional energy at the dipping interface.
220
DEPTH
100
200
300
KM
200 NNA
_l I_
400
600
I
CARNEGIE INSTITUTION
800 1000 KM
1
1 1 1 11
2 1112122 1
24351443398462534321 2
12132235C3331
> 112 2444513 21 V
1 1444622111 1
14A41 1
1 1 31 1 1
1 1111 1222141444 1
12 2 1221 4 412247AT71321111
111 2 221 1421 1
i 11 12 11
11 11
1 1311 1
21 111
1-J1
s] 1 1 1 11
v 3 \ 1 1 1 11311 1 1
1~K 1 11\3 \1 11 111111 1
11 *V 1 \ \ 1 1 113 2 11 1
TV \ \1 11 1 1111 21 1
1 V 1 y 1 V 1 11 222 21 1
-„ 1 \ \ \ 1 111 12 12311
Sty \ 1N 1 31434 31
\$
\ \
%
32311
1
31
X \ \ \ \
20 sec 15 10 5
NNA
1965 Mar 5 SANTIAGO DEL
m = 5.6 H = 555 KM ESTERO
A = 19.6° AzB = 141.2°
ScSp
ScS
10
20 sec
I
Fig. 45. Vertical seismicity cross section and traced seismograms of ScS and ScSp arrivals at
NNA (central Peru) with a circle representing the preferred conversion region. Isacks and Molnar's
( 1971) gently dipping slab model is shown in the inset at the lower left. Their seismicity cross section
is taken over the same region as the one above it, but they include only M > 5 (USCGS, 1961-1967).
fairly well defined position relative to
the dipping seismic plane. In both
South America and Japan (and other
regions as well), the earthquakes in the
dipping seismic plane are at a depth of
about 150 km beneath the line of vol-
canoes. Figure 46 shows the recent vol-
canism in South America as well as
seismic sections through Chile and
Peru. In central Chile, there is a well-
defined dipping seismic plane down to
about 150 km, above which the vol-
canoes occur. In northern Chile, there
is a similar situation, i.e., volcanoes
150 km above the earthquake plane,
except that the seismic plane extends
down to about 300 km. In southern
Peru, volcanoes occur about 150 km
above the dipping seismicity, which ex-
tends down to 250 km.
In central Peru, however, there are
no recent volcanoes. The northern
margin of volcanic activity coincides
with the disappearance of a clear low-Q
zone between the subducting and con-
tinental lithospheres as well as a
poorer delineation of the subducting
plate by the seismicity.
Discussion
For the interaction between the sub-
ducting and continental lithospheres,
three independent observations
suggest that there is a difference be-
tween the southern Peru- Chile region
and that of central and northern Peru
(see Fig. 47).
In the southern regions, there is a
well-developed low-Q zone, presumably
DEPARTMENT OF TERRESTRIAL MAGNETISM
221
100
T C
1 1
200 KM
tDj ^%: =•;'
• X.'.kT '
200
*
<DM
300
-
200KM
1
c;
n
I
1
A
•
100
-
o
Number of
200
-
• •
earthquakes
• 1-2
• 3-4
300
•
o > 5
T C
J L
6
A
200 KM
o
100-
200-
300-
Fig. 46. Recent volcanism and seismicity of western South America. The circles on the map
indicate extinct volcanoes; the triangles, recent volcanism (reproduced from Fig. V.ll, p. 105, in The
Earth's Crust and Mantle by F. A. Vening Meinesz, 1964). The three seismicity sections of CGS
hypocenters on the right side of the figure are taken normal to the trench and are of equal width. T
and C stand for trench and coast, respectively. The asterisk indicates the conversion point for the
ScSp phase recorded by the NNA (Nana) seismograph. The seismograph stations mentioned are
indicated.
indicating temperatures high enough
to cause partial melting, between the
subducting plate and the thick (greater
than 300 km) continental keel. Most of
the seismicity in these regions (see
lower two seismicity sections in Fig.
46) is confined to the subducting plate,
although there is also considerable ac-
tivity in the continental block. Vol-
canoes occur, and the seismic plane is
about 150 km beneath them, as is the
case in other regions such as Japan.
In central Peru, however, the low-Q
zone (sliver of asthenosphere) is small
or absent, and the subducting slab is
not characterized by seismicity below
about 100 km. There is no recent vol-
canism in this region (see Fig. 46).
Note that there are earthquakes at
depths of 150 km, but these occur in-
land far from the subducting slab (as
determined in the studies discussed
above) and are considered to be in the
continental lithosphere.
It appears that the contact of hot and
therefore weak material (characterized
by low values of Q) with the subducting
lithosphere provides the boundary con-
ditions that give rise to a well-defined
seismic plane and the generation of
magma.
We conjecture further that in the
case of an interaction of a subduction
zone with a thick continental litho-
sphere, the cooling effect of the down-
going oceanic lithosphere in the as-
thenosphere between it and the conti-
nental lithosphere can eventually
222
CARNEGIE INSTITUTION
-> '')
Fig. 47. Schematic drawing indicating the key features in the Chile-southern Peru region (left
diagram) and the central Peru region (right diagram). The shaded zone indicates asthenosphere.
Where this exists, the earthquakes (dots) are mainly confined to the subducting slab, and volcanoes
occur above the 150-175 km deep earthquakes. Where asthenosphere is absent, as in central Peru,
volcanoes do not occur and the seismicity is more dispersed.
freeze out that asthenosphere. There is
then direct contact between the sub-
ducting lithosphere and the continen-
tal lithosphere which inhibits vol-
canism and causes diffuse or low seis-
micity. Of the 58°-long subduction zone
south of the equator in western South
America, 25° is already devoid of active
volcanism.
Acknowledgments
This research was supported in part
by National Science Foundation Grant
No. DES 72-01295A02.
References
Barazangi, M., and B. Isacks, Lateral vari-
ations of seismic-wave attenuation in the
upper mantle above the inclined
earthquake zone of the Tonga island arc:
Deep anomaly in the upper mantle, J.
Geophys. Res., 76, 8493-8516, 1971.
Barazangi, M., B. Isacks, J. Oliver, J.
Dubois, and G. Pascal, Descent of litho-
sphere beneath New Hebrides, Tonga-Fiji
and New Zealand: Evidence for detached
slabs, Nature, 242, 98-101, 1973.
Barazangi, M., W. Pennington, and B.
Isacks, Global study of seismic wave at-
tenuation in the upper mantle behind is-
land arcs using „P waves, J. Geophys.
Res., 80, 1079-1092, 1975.
Isacks, B., and M. Barazangi, High-
frequency shear waves guided by a con-
tinuous lithosphere descending beneath
South America, Geophys. J. R. Astron.
Soc.,33, 129-139, 1973.
Isacks, B., and P. Molnar, Distribution of
stresses in the descending lithosphere
from a global survey of focal-mechanism
solutions of mantle earthquakes, Rev.
Geophys. Space Phys., 9, 103-174, 1971.
Isacks, B., J. Oliver, and L. R. Sykes, Seis-
mology and the new global tectonics, J.
Geophys. Res., 73, 5855-5899, 1968.
Katsumata, M., The effect of a seismic zone
upon the transmission of seismic waves,
Kenshinjiho (Q. J. Seismol.), 25, 89-95
(in Japanese), 1960.
Mooney, H. M., Upper mantle inhomo-
geneity beneath New Zealand: Seismic
evidence, J. Geophys. Res., 75, 285-309,
1970.
Okada, H., Geophysical implications of the
phase ScSp on the dipping lithosphere
underthrusting western South America,
Carnegie Inst. Wash. Year Book 73, pp.
1032-1039, 1974.
Okada, H., Ph.D. thesis, Hokkaido Univer-
sity, Sapporo, Japan, in preparation,
1976.
Sacks, I. S., A broad-band large dynamic
range seismograph, in Geophys. Monogr.
10, Am. Geophys. Union, pp. 543-553,
1966.
Sacks, I. S., Q for P-waves in the mantle,
Carnegie Inst. Wash. Year Book 66, pp.
28-29, 1968.
Sacks, I. S., Distribution of absorption of
shear waves in South America and its
tectonic significance, Carnegie Inst.
Wash. Year Book 67, pp. 339-344, 1969.
DEPARTMENT OF TERRESTRIAL MAGNETISM
223
Sacks, I. S., and H. Okada, A comparison of
the anelasticity structure beneath west-
ern South America and Japan, Phys.
Earth Planet. Inter., 9, 211-219, 1974.
Snoke, J. A., and I. S. Sacks, Determination
of the subducting lithosphere boundary
by use of converted phases, Carnegie Inst.
Wash. Year Book 74, pp. 266-273, 1975.
Snoke, J. A., I. S. Sacks, and H. Okada, A
model not requiring continuous litho-
sphere for anomalous high-frequency
arrivals from deep-focus South American
earthquakes, Phys. Earth Planet. Inter.,
9, 199-206, 1974.
Stauder, W., Subduction of the Nazca plate
under Peru as evidenced by focal mecha-
nisms and by seismicity, J. Geophys. Res.,
80, 1053-1064, 1975.
Utsu, T., Regional differences in absorption
of seismic waves in the upper mantle as
inferred from abnormal distributions of
seismic intensities, J. Fac. Sci. Hokkaido
Univ., Ser. 7 (Geophys.), 2, 359-374,
1966.
Vening Meinesz, F. A., The Earth's Crust
and Mantle (Developments in Solid Earth
Geophysics, Vol. 1), Am. Elsevier, New
York, 1964.
Determination of the
Lithosphere- Asthenosphere
Boundary Using Converted
Waves
/. S. Sacks and J. A. Snoke
Introduction
There is considerable evidence for a
thick lithosphere under South
America. A study of surface-wave dis-
persion across the shield area by Alex-
ander (1972) showed an absence of a
low-velocity zone down to at least 200
km, which implies that the lithosphere
extends to at least that depth. Sacks
and Okada (1974) found that there is a
Q reversal at about 350 km depth
under the continent and that the Q at a
given depth above 350 km is approxi-
mately the same as in the downgoing
plate. They concluded that the litho-
sphere extended to about 350 km de-
pth. Also, the seismicity is absent be-
tween depths of about 350 km and 500
km, which is consistent with the theory
SEISMOGRAPH
EARTHQUAKE
Fig. 48. Ray paths of direct S, direct P,
and sp.
that this depth range contains weak
(asthenospheric) material incapable of
supporting the stress concentrations
necessary for earthquakes.
In this report we describe a tech-
nique to determine the depth of the
lithospheric-asthenospheric boundary
by utilizing the fact that the astheno-
sphere probably has a partial melt, so
that there is a significant rigidity con-
trast at the boundary, leading to rela-
tively efficient conversion of incident
shear waves to transmitted compres-
sional waves (Fig. 48; see also Sacks,
1968 and Jordan and Frazer, 1975).
Data
Arrivals intermediate between P and
S were observed at the Carnegie
broadband stations (Sacks, 1966) CUZ
in Peru and TCC in northern Chile,
and at stations employing mechanical,
horizontal seismographs at An-
tofagasta (ANT) in northern Chile, and
in Argentina at La Plata (LPA), Zonda
(ZON), La Quiaca (LQA) and Jujuy
(JUJ). The data are summarized in
Table 13, and station locations and the
epicentral regions of analyzed
earthquakes are shown in Fig. 49.
Examples of the arrivals are shown in
Figs. 50 and 51. The arrivals were ob-
served at epicentral distances of 4° to
13° from events in the depth range
526 -650 km. The locations of epicen-
ters and seismographs are such that it
is impossible that the intermediate
arrivals result from conversions at in-
terfaces related to subduction.
224
CARNEGIE INSTITUTION
TABLE 13.
List of Observed Arrivals between P and S . Locations from
USCGS/NOAA. 57 1
= r arrivaI -
T„-
Date
Depth
Back
Dist.
8T
(d-m-y)
Mag.
(km)
Station
azim.
(deg.)
Comp.
(s)
6-7-59, „
6.75
600
LQA
137
6.06
NS
EW
24
24
6-7-59 (2)
6.9
600
JUJ
125
4.13
NS
EW
22
22-26
LQA
140
5.76
NS
EW
25
22
ZON
53
8.04
NS
21.5
28-12-59
650
LPA
327
8.13
NS
EW
30
19-8-61
7.0
649
ANT
357
12.85
NS
42.5
29-9-62
6.5
575
ZON
46
6.33
EW
NS
20.30
8-12-62
5.9
620
ZON
40
7.36
EW
NS
28
15
9-12-64
5.9
586
ZON
51
6.24
EW
29
5-3-65
5.5
573
ZON
47
6.52
EW
NS
19,29
13-5-65
5.1
589
ZON
21
7.25
NS
24.5
30-7-65
4.5
526
TCC
98
4.17
EW
15
ZON
28
9.78
NS
14.5
20-12-66
5.7
586
ZON
44
7.24
EW
NS
24.5
9-9-67
5.9
578
ZON
53
6.18
NS
24
12-1-72
5.9
580
CUZ
6.67
UD
NS
EW
23
24
Fig. 49. Location map showing seismograph
stations and the epicentral regions (hatched
zones) of analyzed earthquakes. The arrival at
CUZ was for a northern (Peru-Brazil)
earthquake; all observations at other stations
were for southern earthquakes.
Figure 52 has tracings of the
waveforms for the intermediate arrival
(labeled sp) and directs for an event as
recorded at TCC. The similarity of the
two waveforms suggests that the phase
was radiated as a shear wave. The in-
termediate arrival has relatively more
high frequencies than does direct S
(Fig. 50), from which we infer that the
phase traverses the low-Q zone below
350 km (Sacks and Okada, 1974), with
a steeper path through that region
than does direct S. The amplitude of
the arrival is largest on the vertical
component of the seismogram (Fig. 50)
but is in all cases much smaller than
direct S at long periods.
These arrivals were not observable
on WWSSN seismograms because of
the high magnification and limited
dynamic range and frequency response
of these seismographs.
Theory and Interpretation
The only model we have found con-
sistent with the data is that the inter-
DEPARTMENT OF TERRESTRIAL MAGNETISM
, 1 MINUTE
V, 1 SEC
225
V, 2 SEC
E-W, 2 SEC
-S, 2 SEC
||^|^.m>/a>vw^^^^
P sp S.
CUZCO 12 JAN 1972 H = 580
A= 6.1
5.9
Fig. 50. Three-component, filtered P, sp, and S arrivals for an earthquake as recorded by the
Carnegie broadband seismograph CUZ. The upper trace is the vertical component filtered at 1-s
period. The other traces are vertical, E-W (transverse horizontal), and N-S (radial horizontal) fil-
tered at 2-s periods. The sp arrival is strongest on the vertical and weakest on the transverse, as
predicted. The shorter periods ( =£ s) are attenuated in the low-Q region below the lithosphere for S and
sp.
mediate arrivals are sp phases for tinuity, in which case the arrival is ob-
which the S-to-P conversion occurs at served as an immediate precursor to
an approximately horizontal interface directs.
at about 400 km depth (Fig. 48). Previ- The variation of the displacement
ous studies (e.g., Jordan and Fraser, amplitude ratio at sp to S with epicen-
1975) of sp or Sp have considered only tral distance is shown in Fig. 53 (verti-
S-to-P conversion at the Moho discon- cal component) and Fig. 54 (horizontal
a
b
|l| Mil J, i 'II- i ' I'M III ,' '
P sp S
P sp S
LPA 27 DEC 1959
h = 650km
A=8.I3° NS
LQA 6 JULY 1959,
h=600 km m = 6 8
A = 6.06° E W
Hi 1 1 '
I |I)|.J ,|| j| , ,
/If 1:1 'i ; i
P sp
LQA 6 JULY 1959
h=600km m = 6.9
A = 5.76° E W
(2)
Interruption
of energy
supply
ZON 20 DEC 1966
h= 586 km m= 5.7
A =7. 24° EW
Fig. 51. Tracings ofP, sp, and S arrivals for earthquakes as recorded by mechanical seismographs
at ZON, LQA, and LPA. (See Fig. 50 for station locations.) Note that the clear sp arrival on record d
has the same polarity as S.
226
CARNEGIE INSTITUTION
TCC E-W 30 JULY 1965 A = 4.I7
/ ^V VN
Sp (X5)
Fig. 52. Tracings of filtered (at 2.5-s period)
sp and S arrivals for an earthquake as recorded
by the Carnegie broadband station TCC. The
waveforms are very similar, and the polarity is
the same.
radial component). (The sp amplitude
should be zero on the horizontal trans-
verse component.)
The ratios are calculated using ray
theory and include the effects of
geometrical spreading, S-to-S and
S-to-P conversion at the interface (for
enhancements of both 5% and 10% of
the velocities above the interface), re-
lative attenuation below the interface,
and the free-surface conversion of
amplitudes into ground displacement.
For this example of a focal depth of 580
km and a conversion depth of 400 km,
there is a cutoff at about 12°. The rela-
tive amplitude is larger on the vertical
(Fig. 53e) than on the radial (Fig. 54e)
but is always much less than unity.
(Including earthquake radiation-
pattern effects may change the
amplitude ratios somewhat.)
RELATIVE AMPLITUDE OF sp TO S
VERTICAL DISPLACEMENT
FOCUS 580km CONVERSION 400km
RELATIVE AMPLITUDE OF sp TO S
RADIAL DISPLACEMENT
FOCUS 580 km, CONVERSION 400 km
DISTANCE (degrees)
Fig. 53. (a) The transmission coefficient forP
transmitted with an incidents vs. epicentral dis-
tance for sp. (b) and (c) vertical displacement
amplitude for sp and S, respectively, based on
geometrical spreading, (d) Relative attenuation
of S to sp below the conversion depth at 400 km
assuming Q s = 50 and a period of 2.5 s. (e) The
relative amplitude of sp to S for vertical dis-
placement. The focus is at 580 km, and 5% and
10% enhancements for V im _IV i{m+ are included.
DISTANCE (degrees)
Fig. 54. Same as in Fig. 53 but showing the
relative radial displacement amplitudes instead
of the vertical.
The amplitude ratios would be ap-
proximately the same if one assumed
instead a 5% or 10% decrease in the ve-
locities above the interface, and for our
data set, one cannot determine the di-
DEPARTMENT OF TERRESTRIAL MAGNETISM
227
rection of the velocity change at the
interface on the basis of travel time
alone (Fig. 56). It is important
geophysically to know which it is: A
decrease above the interface could
imply that the usual 400-km discon-
tinuity is a sharp discontinuity rather
than the more usually modeled gradual
(over 60-km) velocity increase with
depth, while an increase above the
interface would lead to the interpreta-
tion that the interface is the boundary
between the continental lithosphere
and the underlying asthenosphere.
The polarity of the outgoing P wave
for an S-to-P refraction depends on the
sign of the velocity change at the con-
version interface. The sign of the veloc-
ity change can be determined by com-
paring the waveforms, or even the di-
rections of first-motion, of the sp and
the direct S on radial-component seis-
mograms. Enhanced velocities above
the interface result in iivphase
waveforms and hence in first-motions
in the same direction; enhanced veloci-
ties below the interface result in oppo-
site polarity waveforms and hence in
first-motions in opposite directions.
The first-motions and waveforms from
Figs. 52 and 5 Id are in phase, indicat-
ing that the interface is a velocity re-
versal which we interpret to be the
lithosphere-asthenosphere boundary.
From the plot 8T = T sp -T P vs. epi-
central distance (Fig. 55) we determine
that a one-second increase in ST results
from an increase in focal depth of about
12 km at short epicentral distances, or
about 7 km for larger epicentral dis-
tances. Keeping the focal depth fixed
and decreasing (increasing) the con-
version depth gives similar results for
increases (decreases) in ST.
Figure 56 has curves of 8T vs. epi-
central distance for different velocity-
depth models. Compared with the nor-
mal J-B model, a model with a sharp
increase-with-depth of the velocities
instead of the normal 400-km discon-
tinuity gives essentially identical re-
sults. A model with velocities above the
interface enhanced by 5% (10%) gives
similar 8Ts by up to 3 (6) seconds. For
50r
40-
30-
20-
10
5 10 15
DISTANCE (degrees)
Fig. 55. The calculated difference in sp andP
arrival times is plotted vs. epicentral distance for
a conversion depth at 400 km and focal depths at
526, 580, and 650 km. The long-distance cutoff is
indicated by a dashed line. The J-B model is
used.
1
1
FOCAL
DEPTH km
CONVERSION
DEPTH IS 400 km
Je>50
-
A -
I
/lu
'c/>
.q580_
°
/3
/ //
526
^
1
1
40
30-
uj 20
PERCENT
ENHANCEMENT
CONVERSION DEPTH =400 km
FOCAL DEPTH = 580 km
DISTANCE (degrees)
Fig. 56. Differential travel times as in Fig. 55
for a focal depth of 580 km and a conversion
depth of 400 km for different velocity-depth mod-
els. Indicated are the percentages of the velocity
contrasts for the velocity above the interface rel-
ative to those below.
228
CARNEGIE INSTITUTION
our data these differences are not sig-
nificant.
Figure 57 is a plot of the observed 8T
minus the calculated 8T (using the
normal J-B model) vs. epicentral dis-
tance. It should be noted that the data
include events for focal depths of 526-
650 km and for epicentral distances of
4°-13° and are at a large range of
azimuths. The observations suggest a
probable conversion depth of 400 ±30
km.
An indication of the thickness of the
transition zone can be obtained from
the high-frequency content of the arri-
vals. From Fig. 50 we see that the arri-
val is clearly recorded at periods as
small as 2 seconds, so that to a 10-km
wavelength the discontinuity is sharp.
Hence the transition zone must be only
a few kilometers thick.
Concluding Remarks
Anelasticity and surface-wave
studies, the seismicity and the sp
studies provide conclusive evidence
that there is a thick lithosphere under
South America. Thesp arrivals provide
a high-resolution technique to deter-
mine the lithosphere-asthenosphere
boundary. Our results show that the
bottom of the lithosphere is approxi-
mately horizontal under a substantial
part of the continent; the depth of the
boundary is within a few tens of
kilometers of 400 km; and the transi-
tion zone is at most a few kilometers
thick with a velocity contrast of 5%-
10%.
Preliminary investigations of other
geographical regions indicate that sp
and Sp arrivals exist and that conver-
sion depths correlate with other evi-
dence for the lithosphere-astheno-
sphere boundary in those regions.
Acknowledgments
This research was supported in part
by National Science Foundation Grant
No. DES 72-01295A02.
References
Alexander, S. S., Crust-mantle structure of
shields and their role in global tectonics,
Trans. Am. Geophys. Union, 53, 1043,
1972.
Jordan, T. H., and L. N. Frazer, Crustal and
upper mantle structure from Sp phases,
J. Geophys. Res., 80, 1504-1518, 1975.
Sacks, I. S., A broad-band large dynamic
range seismograph, in Geophys. Monogr.
1 1 1 1
i
1
1 1 1 1 1
INTERFACE DEPTH
8
6
O
-
-
-
___ _- Z= 350 km
4
o
cm
FOCAL DEPTH
a < 570 km
?
O 570-610 km
o
o o
8
A >6IOkm
A_ Z=400km
2
o
o
&
4
D
6
8
i i i i
■ —
o
1
1
~~ — — Z = 450 km
1 1 1 1 1
3 4 5 6 7 8 9 10 II 12 13 14
DISTANCE (degrees)
Fig. 57. Plot of observed 87" minus calculated 8T vs. distance for a conversion depth of 400 km.
Also included are lines corresponding to 8T obse , , ved = 8T calcuIated for conversion depths of 350 and 450
km. An interface depth of about 400 km is preferred. The normal J-B model was used to obtain the
calculated 8T.
DEPARTMENT OF TERRESTRIAL MAGNETISM
229
10, Wash., D. C, Trans. Am. Geophys.
Union, pp. 543-553, 1966.
Sacks, I. S., Converted waves from the
interface at a depth of 410 km. Carnegie
Inst. Wash. Year Book 66, pp. 32-35,
1968.
Sacks, I. S., and H. Okada, A comparison of
the anelasticity structure beneath west-
ern South America and Japan, Phys.
Earth Planet, Inter., 9, 211-219, 1974.
Shallow Seismicity in
Subduction Zones
/. S. Sacks, A. Rodriguez B.*,
J. A. Snoke and A. T. Linde
Introduction
The study described here is one of the
results of long-term collaboration be-
tween this department and the
Characato Observatory of the Univer-
sity of San Agustin in Arequipa.
The seismicity of subduction regions
is characterized by a dipping seismic
plane, some tens of kilometers thick,
which is thought to delineate the upper
surface of the subducting lithospheric
plate. There are no earthquakes in the
asthenosphere on either side of this
subducting plate. Earthquakes do
occur in the continental lithosphere
above the subduction. The seismicity is
not uniform along the subducting
plate; regions of low or zero seismic re-
lease are found.
In this report we describe and inter-
pret the shallow seismicity under
southern Peru using high-resolution
data from a net operated by the Uni-
versity of San Agustin in 1965 (James
et al., 1969). The hypocenters (Figs.
59,60) delineate a 30° dipping Benioff
zone but also indicate intraplate activ-
ity. Of particular interest is the exis-
tence of an aseismic wedge between the
shallower earthquakes in the subduct-
ing lithosphere and those in the conti-
nental lithosphere. We show below
that this aseismicity can be explained
quite simply on the basis of the elastic
forces involved in the tectonics.
* Characato Geophysical Observatory, San
Agustin University, Arequipa, Peru.
Data
During 1965 the Characato Obser-
vatory of the University of San Agustin
operated a net of nine high-sensitivity
seismographs in southern Peru. The lo-
cations of these stations, as well as
some others used for this study, are
shown in Fig. 58.
The method used to locate the
earthquakes recorded by the Arequipa
net is discussed in detail by James et al.
(1969). Stable hypocenters for the local
earthquakes were determined by an it-
erative method utilizing both theP and
the S arrival times. In this method the
origin time is determined indepen-
dently from the spatial location pa-
rameters at stations at which S is ob-
served, and a least-squares iteration
procedure is applied to only the three
spatial parameters.
Figures 59 and 60 show vertical dis-
tributions of shallow (<100 km)
earthquakes located by the Arequipa
net. For comparison, Fig. 61 is a verti-
cal seismicity cross section for the same
geographical area as in Fig. 60, but
using earthquake hypocenters located
■^A
' TRP
V
')
BRAZIL
PERU
t - —
%NNA
HUA
\
/O <g^.
CSP
•
cus j
■y %j
O ?&
ICA
V
s ■%>>.
•
/O <%&.
Ik CRV 0NG
SHP A o
^%s • » /
r\ -
IV :
^&"
ATlS^st
.ARE PUN^<
SGPks,
/ —
°e« ^
Hr~~ AYE ''' *"
v„
u
Fig. 58. Location map of seismograph sta-
tions used in this study. The stations in the
Arequipa net are CUS, PUN, CRV, ONG, SHP,
ATI, SGP, AYE, and ARE. Also used, but not
shown, were ANT (Antofagasta, Chile) and LPB
(La Paz, Bolivia).
230
CARNEGIE INSTITUTION
100
DEPTH
(KM)
200-
T
100
DISTANCE (KM)
c
200
300
1911311 I I
1245412 I I I
I I 122 I 21 I
I 2 1 223 I
24 8433 ill I
I II62ET52
I 3< r 634l I I 2 I I I
I 22154451421 III
M I 1212 313 13
400
1965 SEISMICITY (AREQUIPA NET)
Fig. 59. Vertical cross section of the 1965 seismicity above 100 km located by the Arequipa net for
events inside the rectangle on the map. C and T stand for coast and trench, respectively. Of particular
interest is the aseismic wedge just below the C.
DISTANCE (KM)
o?
T
ioo C
200
300
40
u
i
i
i iii i i
1 12 1
II 12 21 111 11151
14 1 111 1212
1 221 21 1 1 1 1 1 1 III
1 2 3 1 2 1 22 1 1 2 II
1 II 1 1 1 41 121 1
i
100
-
1 1 123253132 5 1
2 2 111 58«9
1 22 24BB72I 1 1
1 133 133 23 1 ! 1 1
II 388332 1 1 1 1
1 i
-
DEPTH
(KM)
-
-
200
-
V '* .
NNA V
15° '■■•tj
mmi!- ■
^iat-
S
o
o> in
30° S |v
WwT
t o
1965 SEISMICITY (AREQUIPA NET)
Fig. 60. Same as Fig. 59 for an adjacent region.
DEPARTMENT OF TERRESTRIAL MAGNETISM
231
by USCGS/NOAA for the period
1964-1974.
In addition, bathymetric data (Getts,
1975), as shown in Fig. 61, indicate a
curvature in the ocean floor on both
sides of the trench.
The Aseismic Wedge
The seismicity cross sections in Figs.
59 and 60 indicate that between the
trench and the west there is an aseis-
mic wedge of about 80 km in lateral
extent and about 50 km in maximum
depth. This feature can also be seen on
the cross section using USCGS/NOAA
locations (Fig. 61), but because of the
smaller data set and less reliable loca-
tions one could not propose the exis-
tence of this wedge on the basis of the
USCGS/NOAA data alone.
A model that is consistent with the
observations is illustrated in Fig. 63. It
is proposed that the hatched area
labeled "low stress zone" in Fig. 63C
corresponds to the aseismic wedge. The
relative plate motion leads to drag at
the upper boundary of the downgoing
plate and compressive stress in the
oceanic and continental plates far from
the subduction (Fig. 63a). (The com-
pressive stresses in these regions are
consistent with the focal mechanisms
found by Stauder, 1975; Mendiguren,
1966). Counteracting the compression
near the subduction zone is tension
caused by the downwarping of the con-
tinental lithosphere due to the drag of
subduction (Fig. 63b). The curvature of
the interacting plates at the trench can
be seen from the bathymetry (Fig. 62).
The tension can be explained by
beam-bending theory (e.g., Sokolnikoff,
1956), according to which, near the
upper surface of a (convexly) bent plate
there is a tension inversely propor-
tional to the radius of curvature. A
superposition of the stresses indicated
in Figs. 63a, 63b results in a low-stress
zone (hatched regions in Fig. 63c).
Earthquakes occur in regions of high or
rapidly varying stress, so that the
lower-stress zones in Fig. 63c are also
seismically quiet regions. That this is a
DISTANCE (KM)
S> |! 100 C 200
300
400
1001-
DEPTH
(KM)
2001-
1
1
1
1
111 II
1 2 1
3 11 1
1 1
1
1
1
1
1
1
t
1
1
1
2
2
1
-
-
0°
■{fizz
15°
NNAV
Tfljjkj?. ;
i
bttij&sjf? -
o
0)
30°
s
in
o
USCGS/NOAA 1964-1974
Fig. 61. Vertical cross section of the seismicity down to 100 km as located by USCGS/NOAA from
1964 to 1974. The events are for the same area as Fig. 60. Even though the data are for a period eleven
times as long as the Arequipa net data, there are only one-fourth the number of events.
232
CARNEGIE INSTITUTION
1 | l | 1
SEA LEVEL
1
1 1 I 1
CONTINENT >,
•
•
1
•
BATHYMETRY
•
•
OFF
•
2
DEPTH
ATICO, SOUTHERN PERU
•
•
(KM)
•
3
•
•
•
•
4
•
•
•
•
•
•
5
— •
•
•
•
•
•
•
•
•
•
6
— •
•
•
•
•
7
ft
i 1 i 1 i
•
•
1
1 1 i 1
500
400
300 200
DISTANCE FROM COAST (KM)
100
Fig. 62. The bathymetry off Atico, southern Peru. The vertical scale is 50 times the horizontal
scale. The curvature of both the oceanic plate (left of trench) and the continental plate is apparent.
COMPRESSION
Fig. 63. Our model for explaining the aseismic wedge, (a) Stresses due to relative plate motion
inferred from fault-plane solutions and frictional drag, (b) Tension due to downwarping of the plates
at the trench due to the drag of subduction. (c) The superpositions of the stresses from a and b result in
low-stress (hatched) regions as shown.
DEPARTMENT OF TERRESTRIAL MAGNETISM
233
global phenomenon can be seen from
high-precision seismicity studies such
as were made in the Aleutians (Eng-
dahl and Sleep, 1976), which also show
a shallow aseismic wedge.
Shallow earthquakes are therefore
not expected between the trench and a
region fairly well inland. In most sub-
duction regions, the coast (C in Figs.
59, 60) is earthquake free.
Acknowledgments
This research was supported in part
by National Science Foundation Grant
No. DES 72-01295A02.
References
ILLUMINATED
REGION, LARGE
AMPLITUDE
ILLUMINATED
REGION, SMALL
AMPLITUDE
Engdahl, R. E., and N. H. Sleep, Seismicity
and stress beneath the central Aleutian
Arc, Trans. Am Geophys. Union, 57, 329,
1976.
Getts, T., Preliminary contour map, Hawaii
Institute of Geophysics, unpublished,
1975.
James, D. E., I. S. Sacks, E. Lazo, L. and P.
Aparicio G., On locating earthquakes
using small networks, Bull. Seis. Soc.
Am., 59, 1201-1212, 1969.
Mendiguren, J. A., Focal mechanism and
pressure axis direction in Central
America and South America, USEE
Bull., 3, 1-20, 1966.
Sokolnikoff, I. S., Mathematical Theory of
Elasticity, McGraw-Hill, New York, 1956.
Stauder, W., Subduction of the Nazca Plate
under Peru as evidenced by focal mecha-
nisms and by seismicity, J. Geophys. Res.,
80, 1053-1064, 1975.
A Model for Laterally
Heterogeneous Velocity
Structure at the Base of the
Mantle
J. A. Snoke and I. S. Sacks
Introduction
Sacks and Beach (1974) showed that
the core phase PKPAB (Fig. 64), which
has near-grazing incidence at the
mantle-core boundary, has an
Fig. 64. Ray paths for the phasePKPAB lead-
ing to large-amplitude (a) and small -amplitude
(b) arrivals due to scattering in the lower mantle
near the core-departure crossing.
amplitude anomaly that correlates
with the regions of the core-mantle
boundary crossings. Regions of high
and normal transmission were iden-
tified, and a correlation was found be-
tween these regions and the isostatic
anomalies in the earth's gravity field
as determined from satellite orbits. We
have used a ray-theory approach to
model regions at the base of the mantle
to determine the characteristics of the
velocity structure that gives rise to
large-amplitude PKPAB arrivals.
It was suggested by Sacks and Beach
that there are two types of regions at
the base of the mantle: regions with
small convection cells (A in Fig. 65)
and regions that are involved in
large-scale (whole mantle) convection
(B in Fig. 65). In this report we de-
scribe, for the A-type region, a
velocity-structure model that leads to
results consistent with the observed
amplitude anomalies.
The Model
Figure 66 shows the velocity-
structure model used for the bottom
150 km of the mantle at either the
core-entry or core-departure region for
234
CARNEGIE INSTITUTION
nTiDluTr)
Fig. 65. Models for the lowermost mantle. (A) Nonconvecting whole mantle. If the whole mantle
does not convect, the lower viscosity region (Sacks, 1972) at its base probably will, the convection cell
being bounded at the top by static material with a higher viscosity. (B) Whole-mantle convection (or
plumes).
MANTLE
VELOCITY MODEL AT
BASE OF MANTLE
V(X,Z)=V (Z)
He
I + COS(2tt-^ +</>)] j
Fig. 66. The velocity model for the lowest 150 km of the mantle used in this study. x u is fixed
relative to the source. The minimum velocity at a given depth z is V (z), and the maximum is V (z)
(\+A).
the core phase PKPAB , the phase that
has the amplitude anomaly when it
crosses a type-A region. Figure 67 is a
comparison of our model and the struc-
ture inferred by the small-scale convec-
tion cells.
Our model superimposes a cosine
lateral (x) variation on a normal
velocity-depth model (the B 1 model
developed by Jordan and Anderson,
1974). In our model the lateral varia-
tion is independent of depth in the low-
ermost 150 km of the mantle. The pa-
rameter jc„ is kept fixed relative to the
source, so there are three variables in
the model (Fig. 66): A is the maximum
deviation of the velocity from the nor-
mal value at a given depth (all devia-
tions are positive); L, the "wavelength"
of the cosine, is a measure of the lateral
extent of a cell; and $, the "phase" of
the cosine, gives the lateral position of
the cells relative to the source. Only
two-dimensional structure was consid-
ered, and only great-circle paths were
calculated directly. The simulation of
the third dimension is discussed below.
For each model run A , L, and $ were
fixed for either a core-entry or core-
departure crossing. Then ray paths
were calculated to include all PKPAB
arrivals with epicentral distances
greater than 158°. Relative amplitudes
were calculated by comparing the
DEPARTMENT OF TERRESTRIAL MAGNETISM
235
MANTLE
Fig. 67. Comparison of our velocity model and the convection cells it is intended to simulate. The
velocity variations in the model are independent of the depth.
change in epicentral distance for a
given change in departure angle from
the source (Fig. 64). The two-dimen-
sional ray-tracing routine is based on a
method developed by Jacob (1970).
In a laterally homogeneous medium,
the ray parameter p = dT/dA is con-
stant along a ray path, where T is the
travel time and A the epicentral dis-
tance between any two points along the
ray path. Lateral heterogeneities re-
sult in variations in p. Rays traveling
to a single point at the earth's surface
can come from a region at the base of
the mantle of extent d, which can be
determined if the variation in p is
known. We simulate off-great-circle
paths by which a ray travels from
earthquake -to seismograph by assum-
ing that the deviations off the great
circle are similar to those along it —
that is, by superimposing, with a suit-
able weighting, runs of a range in 4> of
8</> = 360d/L.
The lateral position of the cells rela-
tive to the source is probably not de-
terminable, so we average over $.
Hence our model is not strictly deter-
ministic, and the comparison with data
must be on a probabilistic or statistical
basis.
The observed AB amplitudes are de-
fined relative to the PKBDF amplitude
for the same earthquake-seismograph
pair (Fig. 68). The DF branch has ap-
proximately normal incidence at both
core-mantle crossings, which makes it
far less sensitive than the near-grazing
AB rays to any lateral velocity struc-
ture at the base of the mantle. Our
criterion for large AB amplitude is that
it be 1.5 times the size of the corre-
sponding DF amplitude. Our compari-
son with the data will be on the basis of
the percentage of large amplitudes and
the ability of the model to predict the
observed frequency dependence of the
amplitude anomaly.
Results
For each run, the lateral heterogen-
eities cause multiple arrivals at most
epicentral distances (Fig. 69). Suffi-
ciently small steps in departure angle
were taken to ensure that no aliasing
occurred. The difference in arrival
times of the mutliple arrivals is small
enough (<0.2 s for a 1.5% heterogene-
ity model) that, for periods considered,
the arrivals can be added coherently.
We considered models with L = 40 km,
150 km, and 600 km; and A = 0.008,
0.015, and 0.03. The analysis will be
described in detail for the model we
found to be most successful: A = 0.015
and L = 150 km. (The rejected models
and the criteria for their rejection will
be discussed later.) So far we have
analyzed in detail applications of the
model to only the core-departure re-
gion.
236
DF' AB
SP 1
DF
CARNEGIE INSTITUTION
CAR 24 MARCH 1970 A = 163°
AB
PTO 7 OCTOBER 1966 A = 160°
-^V\/y\^A/*^^•^^•v^A-^»«'■^'W\/-^A/*~■^^^^
LP ,\
MAL 7 OCTOBER 1966 A = 164°
DF ¥ U AB
Fig. 68. Tracings oiPKPAB and PPBDF arrivals from long-period and short-period seismograms
illustrating the possible frequency dependences for large A5-amplitude anomalies. For the top pair
the AB amplitude is large on only the short-period record; for the middle pair it is large on both, and
for the bottom it is large on only the long-period record. For an explanation in the context of our
model, see text and Fig. 74.
l50XI50Km,l.5%, 0=0°
to
[V 1234.5 - (A- 159) X 4. \\
<L>
to
TRAVEL TIME
O
ho
•
/
-
Q
UJ
S oi
C .08
t i-fif
-
•0 6
-
.0 4
-
.02
1 1
—
159.0
159.5
EPICENTRAL DISTANCE (degrees)
160.0
Fig. 69. Plot of part of the travel-time curve for a model run with L = 150 km, A = 0.015, and (/> =
0°. The circles are model-derived points, and the lines connecting the circles indicate the continuous
variation in the input ray parameter/} = clT/dA. Note that the scatter in arrival time is <0.2 sec.
DEPARTMENT OF TERRESTRIAL MAGNETISM
237
Figure 70 contains a summary of the
model data. Plotted are the large-
amplitude regions at epicentral dis-
tances for {/> ranging from 0° to 330° in
30° steps. For each run (a fixed (f>) one
sees there are brightly illuminated re-
gions as well as small-amplitude re-
gions, as is required by energy conser-
vation. Some of those bright regions
are less than 0.1°; others are as large as
a half degree. For different 0s the pat-
terns are similar, but the position of
the bright regions shifts. For these
runs the standard deviation in the
change in the ray parameter/? was 0.04
s/degree, which implies a source region
at the base of the mantle of extent d ~
200 km. Following our technique for
simulating off-great-circle paths, this
results in a range in cf> to be superim-
posed of ±240°. The resulting model
probability for large amplitudes is
34%, and this result is insensitive to
the central value of (/>.
In the data analysis, event-
seismograph pairs were selected so
that the data set included a large area
at the core-entry crossing for the rays
for a given core-departure region (Fig.
71). The data therefore include rays
with traversal regions of both type A
and type B on the core-entry side. As-
suming that the model results for the
core-entry side will be similar to our
results for the core-departure side and
that the two types of regions have ap-
proximately equal probability, we get
an additional 11% in expected large-
amplitude arrivals and thus an overall
expectation of 45%. The observed
large-amplitude-arrival probability for
type A regions is 48% (Fig. 72).
A model with L = 600 km and A =
0.015 led to a 7% probability of large
amplitude arrivals, and a model with L
= 40 km and A = 0.015 led to a proba-
bility approximately the same as the
model withL = 150 km and A = 0.015.
The probability for L = 600 km can be
increased by increasing A , but in order
to get results comparable with the
other model, A must be large enough to
cause measurable shifts in the travel
times, which are not observed. Hence
we reject L = 600 km.
The other two models give different
predictions for the size of the illumi-
nated (large-amplitude) regions on the
earth's surface (Fig. 73). The L = 150
km model is capable of explaining the
300
"^ 240
a>
% 180
^Ql
120-
60-
□
am
II
I I
□
1
D
II
a
DD
II
I
1 1
□
II
I
D
D I
I II I
150 XI50Km, 1.57.
Large amp. probability 34%
S.D. A«l.8°
S.D. dT/dA= .04 sec /degree
155
160
EPICENTRAL DISTANCE (degrees)
165
Fig. 70. Results of model runs for A = 0.015 and L = 150 km and different values of (/>. Large-
amplitude arrival regions on the earth's surface are indicated by rectangles. Regions smaller than
0.1° are filled rectangles. See the text for the method of determining the large-amplitude-arrival
probability.
238
CARNEGIE INSTITUTION
Fig. 71. Technique used to resolve the ambiguity of the two core crossings. The dotted lines
indicate areas at the base of the mantle whose transmission properties are being investigated. L
indicates low-amplitude transmission, and H indicates high amplitudes. Circles are seismographs;
asterisks, earthquakes; full lines show mantle paths; and broken lines, paths in the core. The average
amplitude for theL region sampled is {LL + LH + LL + LH)/4. If the adjacent// region was sampled,
the average would be (HL + HH + HL + HH)/4, which is clearly larger.
different types of observations (Fig. 68;
see also Fig. 74): (1) where the
amplitude of the AB is large on the
short-period but not on the long-period
seismograms (because the extent of the
illuminated region is small compared
to the wavelengths for which the long-
period seismometer is sensitive), and
(2) where the amplitude is large on
both short-period and long-period in-
struments (because the extent of the il-
luminated region is above the
threshold for the long-period seis-
mometer). For the L = 40 km model
there are no illuminated regions larger
than 0.1° in length (Fig. 73). We can
therefore rule out the L = 40 km model
because it cannot explain the observa-
tions for which the AB arrivals are
large on both long-period and short-
period seismograms (e.g., Fig. 68).
Model runs for the PKPDF branch
using the same parameterization as for
the PKPAB branch produced results
that were essentially the same as for
the laterally homogeneous {A = 0)
case.
Concluding Remarks
We have shown, therefore, that in
some regions at the base of the mantle
cells of 150 km x 150 km with a
maximum lateral velocity variation of
1.5% can explain the observed
amplitude anomaly of PKPAB arrivals.
The possibility of convection occurring
is proportional to the cube of the cell
size, so it is encouraging for the convec-
tion model that smaller cells are not
required to fit the seismic data. The
limits on the cell size and velocity vari-
ations obtained from this study impose
constraints that will be invaluable in
studies of convection and heat flow at
the core-mantle boundary.
DEPARTMENT OF TERRESTRIAL MAGNETISM
239
SIZE OF AB ARRIVAL
Fig. 72. Number versus size of the PKBAB
amplitude arrival for a type-A (small-scale con-
vection) region for the core-departure crossing of
the PKPAB rays. The percentage of large arri-
vals is 48.
Acknowledgments
This research was supported in part
by National Science Foundation Grant
No. DES 72-01295A02.
References
Jacob, K. H., Three-dimensional seismic
ray tracing in a laterally heterogeneous
spherical earth, J. Geophys. Res., 75,
6675-6689, 1970.
Jordan, T. H. and D. L. Anderson, Earth
structure from free oscillations and travel
times, Geophys. J. R. Astron. Soc, 36,
411-459, 1974.
Sacks, I. S., Viscosity and Q, Carnegie Inst.
Wash. Year Book 71, pp. 227-229, 1972.
Sacks, I. S. and L. Beach, Lateral
heterogeneity at the base of the
mantle — an indication of whole mantle
convection, Carnegie Inst. Wash. Year
Book 73, pp. 1020-1032, 1974.
50
I
1.5% Heterogeneity
Height ofcell = l50Km
30
Short period seism.
Long period seism.
-
10
\ 150 X 150 Km
-
5
-
-
3
1
1
40X150 Km \
0.001
0.01 0.1
ILLUMINATED DISTANCE (degrees)
1.0
Fig. 73. Number versus size of the illuminated (large-amplitude) region on the earth's surface for
L = 150 km and L = 40 km. Also shown are the lower limits of the sensitivity of the WWSSN
long-period and short-period seismometers for an arrival to be recorded as large. The fact that there
are observed large amplitudes on long-period seismograms rules out L = 40 km.
240
CARNEGIE INSTITUTION
Fig. 74. Explanation for the observed frequency dependence of the large AB amplitudes (see Fig.
68 for examples). A large amplitude on only short-period records (A,) occurs for a triplication over an
epicentral distance range of less than 0.1°. A large amplitude on only the long-period record (A 2 )
results from diffractive scattering from the caustic. Such arrivals were not considered in this study.
Large amplitudes on both short-period and long-period records (A,) result from a triplication over an
epicentral distance range greater than 0.1° in extent.
Single Site Phase Velocity
Determinations
/. S. Sacks, J. A. Snoke, J. R. Evans, G. C. P.
King* and R. J. Beavan*
The lithosphere beneath oceans is
generally less than 120 km thick. Evi-
dence is mounting that beneath large,
old continental areas, such as South
America, the lithosphere is very much
thicker. Anelasticity studies by Sacks
and Okada (Year Book 72, pp. 226-233)
and the interpretation of converted sp
phases (Sacks and Snoke, this Year
Book) indicate that the lithosphere be-
neath South America must be about
400 km thick and that it overlies as-
thenosphere which has much higher
absorption (lower Q).
A method to determine subcrustal
velocity structure utilizes surface-wave
phase velocities (see Fig. 75). Phase ve-
locities are traditionally obtained by
using sets of widely spaced seismome-
ters which record simultaneously. The
velocity determination follows then
from measurements of the travel times
of different frequencies over known
distances. For low-frequency (long-
* Department of Geodesy and Geophysics,
Cambridge University, Cambridge, England.
wavelength) waves, the separation of
the instruments must be large to per-
mit satisfactory resolution, and even at
shorter wave lengths, large spacing is
an advantage as it prevents velocity es-
timates being contaminated by multi-
pathing resulting from lateral refrac-
tion and diffraction. The best results
for phase velocity using these methods
are obtained only when two or more
stations lie close to a greater circle in-
cluding the seismic epicenter (Brune
and Dorman, 1963).
The lithosphere asthenosphere
boundary is characterized by a velocity
reversal and beneath large, old conti-
nents this boundary may be as deep as
400 km (see Sacks and Snoke, this Year
Book). Because of the long-period limi-
tations in the determination of phase
velocities by the technique described
above, Alexander (1972) was able to
show only that there is no velocity re-
versal under the shield area in South
America above 200 km.
We describe a method whereby it is,
in principle, possible to determine the
phase velocity of any seismic wave
from the simultaneous measurement of
the wave amplitude (or its time deriva-
tive) and its spatial derivative (strain
or tilt) at the same site. Mikumo and
DEPARTMENT OF TERRESTRIAL MAGNETISM
PHASE VELOCITY DETERMINATION
241
TWO STATION METHOD
SEISMOMETERS 4 AT A AND B
C(T)
t B -t.
SINGLE SITE METHOD
SEISMOMETER
AND STRAINMETER
AT SINGLE SITE
C(T)-
DISPLACEMEN T
STRAIN
Fig. 75. Sketch illustrating the two methods for determining the phase velocity: the single-site
method discussed in this report and the traditional two-site method. T, the period, = 2tt/<d.
Aki (1964) have previously used this
technique for the combined analysis of
records from a horizontal seismometer
and a horizontal strainmeter at a
single station. For the five earthquakes
studied they found good agreement be-
tween measured and predicted phase
velocities for body waves and in some
cases for surface waves.
Rodgers (1968) has shown that for
long periods (>100 s) horizontal accel-
eration and tilt cannot easily be sepa-
rated instrumentally, and recent work
by King et al. (1976) indicates that
near-station crustal inhomogeneity
may give rise to strain-strain and
strain-tilt interactions. Thus, records
from long-period horizontal strain-
meters are heavily contaminated by
tilt and strain. Records written by ver-
tical seismometers do not suffer this
defect. We rederived (Sacks et al., 1976)
the principle of single-site velocity de-
termination, taking into account the
strain-coupling effects noted by King et
al. and using vertical acceleration. Our
discussion is presented for Rayleigh
waves and assumes that the ellipticity
of particle motion at the surface is in-
dependent of frequency. (The validity
of this assumption may be tested if
there is a site-calibrated horizontal
seismometer at the same site.)
In this report we present and discuss
a method for phase velocity determina-
tion employing a vertical seismometer
and a volume strainmeter. The deriva-
tion for this case and for the more gen-
eral case of a vertical seismometer and
a linear strainmeter are to be found in
Sacks et al. (1976).
From Sacks et al. (1976) the phase
velocity c as a function of angular fre-
quency, co, can be written
c(co) = YW(co)/co,
(1)
where Y is a constant (assuming the
ellipticity to be independent of fre-
quency) and
W(to) = u :i (co)/D(co),
(2)
where u :i is the vertical displacement
and D the dilatation. The ~ indicates
that the quantity is averaged over
areas of extent large compared to the
size of the local heterogeneities but less
than or equal to the wavelengths of
interest. If the region^were uniformly
layered, u A = u :i and D = D so that W
from Equation 2 would be given in
measured quantities. However, if there
are local heterogeneities, King et al.
(1976) have shown that, to a good ap-
proximation
and
u-> (co) = u : , (co) (3)
D(d,co)=A(d)D(co), (4)
242
CARNEGIE INSTITUTION
where 6 is the back-azimuth of the
Rayleigh wave. Hence from Equations
2-4, Equation 1 can be rewritten in
terms of the measured quantities u :i
and/):
where
c((o) = YA($)W(e,ai)/(o, (5)
W(d, o)) = u :i ((o)/D(6, w). (6)
There is therefore an azimuthally
dependent factor Aid) in the expression
for the phase velocity due to the pre-
sence of local heterogeneities. Figure
76 is a sketch of log 10 c(^>) vs. co for dif-
ferent azimuths based on Equations 5
Log, [EM]
Fig. 76. Sketch to show the expected rela-
tionship of log,„[c(w)] versus w for Rayleigh
waves arriving from three different directions 0, ,
6>, and :i . The curves are shifted vertically rela-
tive to each other by the presence of local
heterogeneities, but their shape remains un-
changed.
and 6. If the region is uniformly
layered the curves for different
azimuths should superimpose. The pre-
sence of local heterogeneities, on the
other hand, would result in the curves
being displaced from one another.
Hence determining the phase velocity
for Rayleigh waves from a number of
different azimuths allows an estimate
of the local heterogeneity. (An alterna-
tive technique for determining the site
effects using tidal observations was de-
scribed by King et al., 1976.)
The shape of the curves in Fig. 76 is
independent of Aid). By normalizing
these curves at some frequency and av-
eraging over determinations for a
number of different azimuths, it should
be possible to reduce the effects of
signal-generated noise in the determi-
nation of the spectral shape of the
phase velocity. If c(&>) can be deter-
mined at any frequency o> by other
methods (such as from crustal refrac-
tion), Equation 5 can be rewritten as
c(a>) = c(w o )W(0,w)/W(0,a>„)
(7)
Hence the phase velocity can be deter-
mined for all frequencies without
knowing the ellipticity or Aid).
Single-site phase velocity determi-
nations have two main advantages:
There is no inherent limitation in the
maximum period such as is caused by
finite station spacing using the tradi-
tional techniques; and phase velocities
can be averaged for any azimuth,
which will increase the precision of the
velocity determinations, particularly
at longer periods. This technique will
be particularly useful in studies of the
structure of continental lithospheres
with thicknesses up to 400 km.
The implementation of this tech-
nique requires seismometers (ac-
celerometers) and strainmeters with
high sensitivity at long (>400 s)
periods. Suitable strainmeters have al-
ready been developed at the Depart-
ment of Terrestrial Magnetism (see
Year Book 68, pp. 448-452; Year Book
69, pp. 426-430; Year Book 70, pp.
336-340; Year Book 71, pp. 321-325;
Year Book 72, p. 326; Year Book 73,
1058-1060; Year Book 74, pp. 287-
291). A suitable seismometer is being
developed.
Acknowledgments
This research was supported in part
by National Science Foundation Grant
No. DES 72-01295A02.
References
Alexander, S. S., Crust-mantle structure of
shields and their role in global tectonics,
EOS. Trans. Am Geophys. Union, 53,
1043, 1972.
DEPARTMENT OF TERRESTRIAL MAGNETISM
243
Brune, J. N., and J. Dorman, Seismic waves
and earth structure in the Canadian
Shield, Bull. Seis. Soc. Am., 53 (1), 167-
209, 1963.
King, G. C. P., W. Zurn, R. Evans, and D.
Emter, Site correction for long-period
seismometers, tiltmeters, and strain-
meters, Geophys. J. Roy. Astron. Soc,
44, 405-411, 1976.
Mikumo, T., and K. Aki, Determination of
local phase velocity by intercomparison of
seismograms from strain and pendulum
instruments, J. Geophys. Res., 69, 721-
731, 1964.
Rodgers, P. W., The response of the horizon-
tal pendulum seismometer to Rayleigh
and Love waves, tilt and free oscillations
of the earth, Bull, Seis, Soc. Am., 58,
1364-1406, 1968.
Sacks, I. S., R. Evans, G. King, J. Beavan,
and J. A. Snoke, Single site phase veloc-
ity measurement, to appear in Geophys.
J. Roy. Astron. Soc., 1976.
Maximum Entropy Spectral
Analysis
C. Seber, D. E. James,
J. A. Snoke, and A. T. Linde
Introduction
Spectral analysis of seismic waves is
used to investigate a wide range of
problems in seismology. This is be-
cause a great deal of information con-
tained in a seismogram is not readily
observable in the time domain but can
be observed in the frequency domain.
Frequency domain information for
both body waves and surface waves had
been used by us previously to study
earthquake mechanisms and earth
structure (Year Book 69, pp. 447-464,
and Year Book 70, pp. 343-344). For
body waves generated by earthquakes,
we calculated spectral amplitudes as
functions of frequency to investigate
the nature of rupture processes in the
earth (e.g., fault length, speed of rup-
ture propagation along the fault, fault
offset); for dispersed surface waves,
period was plotted vs. time to obtain
dispersion curves that yielded informa-
tion on the velocity structure of the
earth's crust and the upper mantle be-
neath the central Andes.
In nearly all previous work, both by
us and by others, frequency domain in-
formation was obtained by Fourier
analysis. The advantages of the
Fourier technique are many, including
the fact that it is completely under-
stood theoretically and has a fast com-
puter algorithm. Because of these obvi-
ous advantages, most workers have
been content to ignore, or at least tol-
erate, the manifest disadvantages of
the technique, notably that it gives
poor resolution of spectral peaks, espe-
cially at long periods, and that it
makes arbitrary and predetermined
assumptions about the behavior of the
signal outside the interval in which it
is being analyzed. To understand these
limitations better, we must first recog-
nize that any block of data is by neces-
sity only a finite subset of an infinite
data set. Therefore, any scheme of
spectral analysis necessarily results
only in a "spectral estimate." To keep
the error of this spectral estimate
within tolerable limits using Fourier
analysis, it is usually necessary that
the length of data be many times
longer than the greatest period of
interest. Even when this is possible,
however — and it commonly is not —
Fourier analysis often cannot yield
satisfactory results because it assumes
that the signal outside the data block is
either zero or a periodic duplication of
the data within the block to be
analyzed. These inherent weaknesses
in Fourier techniques led Burg and
others in the late 1960s to discard such
convenient devices and to seek a more
flexible approach to spectral analysis.
(For a more complete and particularly
lucid summary of the rationale for dis-
carding Fourier techniques and apply-
ing the maximum entropy methods in-
stead, see McGee 1969.)
Maximum entropy spectral analysis
as formulated by Burg (1967, 1968) re-
sults from applying entropy consid-
erations to spectral determinations so
that information is carried from the
time domain (e.g., seismogram) to the
244
CARNEGIE INSTITUTION
frequency domain with a minimum of
distortion. The method is particularly
powerful in that it avoids the Fourier
assumptions about the nature of the
signal outside the analyzed region; in-
stead, no a priori assumption is made,
and an estimate of the signal outside
the known interval is based only upon
information (entropy) extracted from
the data themselves. (Entropy is used
in the sense of statistical entropy as de-
fined by Shannon and Weaver [1949]
and may be considered to be equivalent
to information. Since we wish the
spectral estimate to exhibit as much in-
formation as is fully consistent with
the data and no more, we state that
[McGee, 1969]: "The most reasonable
spectral estimate for any set of data is
that estimate which exhibits
maximum entropy and yet fully agrees
with the available data."). The
maximum entropy method (MEM) has
been shown to give improved spectral
resolution compared to Fourier
analysis (FA) for many kinds of sig-
nals.
The theoretical basis for MEM is
given by Burg (1967, 1968), McGee
(1969), Claerbout (1976), Seber (1976),
and many others. A recent review of
the technique is given by Ulrych and
Bishop (1975), who also evaluate the
optimum filter length criterion of
Akaike (1969, 1970). Our computa-
tional methods are based on algorithms
given by Ulrych and Bishop.
Most previously published work on
MEM has involved simple synthetic
signals such as sine waves or superim-
posed sine waves with various amounts
of added random noise. Results from
these studies have shown that: (1)
MEM produces highly peaked spectra
compared to FA for short record seg-
ments (one half to two cycles) of con-
tinuous sine waves, giving frequencies
that are correct to within a few percent.
The addition of noise to the signals
does not significantly degrade the re-
sult. The largest single effect on calcu-
lated frequency appears to be the start-
ing phase of the input signal (Chen and
Stegen, 1974). (2) For superimposed
sinusoids of different frequencies,
MEM resolves the frequencies far bet-
ter than FA (Fig. 77). If the sinusoids
have equal amplitude, the power
spectra should have equal areas under
the curves. MEM appears to err in this
regard by as much as a factor of 2. The
theoretical reason for this is not
known, and as a consequence
amplitude instability has been a major
reason MEM has not been more widely
used for spectral analysis. We note,
however, that FA produces even
greater errors for these simple signals.
It is the purpose of this study to test
MEM on synthetic seismic signals to
evaluate the applicability of the
method to problems in seismology. We
have analyzed both body waves and
dispersed surface waves and obtained
results that show sufficient promise to
warrant further investigation.
Synthetic Seismic Signals
Body waves. Difficulties in the
Fourier frequency analysis of body
waves often arise because the signal of
interest may be affected by either ear-
lier or later arriving phases. The prob-
lem is to determine the spectral con-
tent of the complete signal, uncon-
taminated by other arrivals. Because
MEM works well on short pieces of in-
formation relative to Fourier tech-
niques, it might prove a more effective
tool for extracting frequency informa-
tion from signals suffering interference
from earlier or later arrivals.
Moreover, the prediction potential of
MEM might allow one to predict fre-
quency content of the interfering sig-
nal so that it could be subtracted from
the signal of interest. Another tech-
nique that might be considered (Ulrych
et al. , 1973) is using MEM to extend the
given short signal to a length where
Fourier analysis can be used success-
fully. This seems to have been done
with considerable success. Where
phase information is of interest this
would be especially useful, since phase
DEPARTMENT OF TERRESTRIAL MAGNETISM
1.0
245
Ld
O
Q_
Q
LU
N
o
0.8
0.6
0.Q
00 40 80 120 160 200 24.0
TIME (DATA INTERVALS)
^
-- . I--.
0.2
0.3
0.4
0.5
FREQUENCY (CYCLES/DATA INTERVAL
The Fourier power spectrum is given
by
PM = n -^-—{l-e-"[2a+b)cos
bco +2b(o sin bw] + e~ 2b [(1 + b) 2
+ b 2 or}
Fig. 77. Burg power spectrum and periodogram of signal shown in inset. Input signal consists of
20 points at 1-s intervals of a time series made up of three superimposed equiamplitude sine waves
with frequencies of 0.04, 0.08, and 0.12 Hz. Adapted from Ulrychet al., 1973.
information is available in the Fourier
transform but is lost in MEM.
Simple radiated pulse. Brune (1970)
gives an expression for the radiated
pulse for a simple earthquake model.
From many studies using real data, it
has been shown that the shape of the
displacement amplitude in the fre-
quency domain for an earthquake is
reasonably well approximated by
Brune's empirically derived model.
This model has simple analytical ex-
pressions in both the frequency and
time domains and provides a
straightforward comparison of the re-
lative accuracy of FA and MEM for
signals of the kind encountered in
seismology.
Brune's model is defined by a func-
tion, f(t), for a finite time interval such
that fit) =te~'H(t) where
H(t) =
t <0
1 t>0
where b is the length of the signal in
the time domain. No truncation corre-
sponds to the limit 6^> °°.
Figure 78 shows a Brune's model
pulse with various points of truncation.
The maximum entropy spectrum
agrees very closely with the theoretical
spectrum (see Fig. 79). Truncating the
signal on the left (reducing the number
of zeros) had no significant effect on the
spectrum. With just one zero value pre-
ceding the signal onset, MEM was still
able to detect the discontinuity and
give a good spectrum.
246
CARNEGIE INSTITUTION
The effects of successive truncations
on the right side of the signal are
shown in Fig. 78. For a -1 to 4 second
record, MEM was significantly better
than Fourier analysis. A signal trun-
cated at point 61 (see Fig. 78) gives a
MEM spectrum that compares very fa-
vorably with theory. Similarly, signals
from points 15 to 31 (5 s) and from
points 15 to 25 (3 s) also give very good
spectra, although they are not as accu-
rate at higher frequencies. The 15-21
(2 s) segment produces a spectrum that
deviates significantly from the spec-
trum of the full signal. It was observed
that the zero frequency amplitude was
inversely proportional to the length of
the window. This seems to indicate a
high degree of amplitude stability.
The MEM results can be compared to
those obtained using Fourier analysis.
At a 5-s truncation, there is a strong
effect on the spectrum which progres-
sively worsens with increasing severity
of truncation (see Fig. 78).
We conclude that the MEM gives a
very good estimate of the true spectral
content even if the pulse is severely
truncated. This result is encouraging
and indicates that MEM may be useful
in estimating the spectral content of
truncated seismic signals.
Sine wave convolved with instrument
response. Linde and Sacks (1971) have
described the spurious spectral
amplitudes in the long-period part of
the spectrum that result from having
to analyze long-period body-wave arri-
vals of essentially one cycle. Studying a
sine wave convolved with instrument
response, they found that increasingly
severe truncations resulted in increas-
ing emphasis of long-period ampli-
tudes, even though the amplitude of
the truncated part of the signal was
small (see Fig. 79).
Q_
<
O
LlI
M
0.01
0.001
FREQUENCY
Fig. 78. Maximum entropy and Fourier power spectra of Brune's model pulse (see inset) for
various truncation points indicated in inset. Point intervals over which signal has been analyzed are
indicated on the figure. MEM andFA denote maximum entropy method and Fourier analysis, respec-
tively.
DEPARTMENT OF TERRESTRIAL MAGNETISM
247
TRUNCATION POINTS
60
I
>/oOF MAX.
TIME (SEC)
I CYCLE OF 20 SEC PERIOD, CONVOLVED
WITH 30 SEC SEISMOMETER-100 SEC
GALVANOMETER INSTRUMENT
/,•/•'" \ \ — 78 SEC
// / \ \— — 50 SEC
/ // '—-38 SEC
INPUT SIGNAL SPECTRUM
/ (.78.5 SEC)
I
200 100 50
30 20
PERIOD (SEC)
EFFECT OF TRUNCATION
NOISELESS, BOXCAR WINDOW 20SEC SINGLE CYCLE
Fig. 79. Normalized maximum entropy power spectra (dashed lines) and Fourier amplitude
spectra (solid lines) showing the effect of truncation in the time domain (upper figure) of a noiseless
1-cycle 20-s sine wave convolved with a 30-s seismometer, 100-s galvanometer instrument. Successive
truncation at shorter and shorter times drastically affects the long-period end of the Fourier spectrum
but results mostly in a shift in the peak frequency for the maximum entropy spectra. The input signal
spectrum is the theoretical Fourier spectrum.
When this signal is essentially of in-
finite length, the area under the curve
is zero (i.e., the zero frequency term
will be zero) but truncation results in a
positive area (i.e., positive zero-
frequency term). Erroneous low-
frequency amplitudes are further
magnified when multiplied by the
transform of the instrument response,
which is large at very low frequencies.
It was hoped that MEM would not
produce these spurious effects for the
long-period part of the spectrum. As
can be seen in Fig. 79, MEM makes a
good estimate of the frequency of the
convolved sine wave but does not pro-
duce zero amplitude at very long
periods. Hence, multiplication by the
instrument correction causes an er-
roneous rise in amplitude at the long-
period end of the spectrum. We note,
however, that the amount of spurious
long-period signal is not due to trunca-
tion effects but seems to remain
roughly constant regardless of the
amount of truncation. We are unable to
248
CARNEGIE INSTITUTION
explain these spurious amplitudes at
long periods, since for single sine
waves MEM has no difficulty in recog-
nizing zero area under the curve. As a
further test of MEM on more complex
signals with zero spectral amplitude at
zero frequency, we also analyzed a
double Brune pulse. The double pulse
has the property that its Fourier trans-
form at zero frequency is zero. The
function fit) is given by fit) — {te~' —
a 2 te at \ Hit). It begins discontinu-
ously in the time domain and dies off
exponentially as t — » °°. Once again
MEM gives similar incorrect results in
the low frequency range, tending to a
nonzero amplitude plateau rather than
to zero.
Surface waves. All modern analytical
techniques used to obtain dispersion
curves for surface waves require that
some form of filtering (in either the
time or the frequency domain) be
applied to the record. These techniques
are generally satisfactory for short
periods but are incapable of high reso-
lution at long periods. This is to say
that as the period of a given wavelet
increases it becomes more difficult to
determine its exact "arrival time."
Some of this difficulty arises from an
inherent uncertainty of measuring the
time of arrival to an accuracy much
greater than the time scale of a single
cycle, but it seems likely that part of
the problem lies with the analytical
techniques. Because MEM can give
vastly improved results over Fourier
analysis for short segments of har-
monic waves, it is appropriate to inves-
tigate whether or not it can provide
better time resolution for the arrival of
long periods over conventional tech-
niques.
To test the performance of MEM on
dispersed waves, we generated a syn-
thetic signal that was linearly disper-
sive with a flat amplitude spectrum be-
tween 0.078 and 0.156 Hz. The spec-
trum was cosine-tapered over 20
Fourier harmonics outside that fre-
quency interval. Various window
lengths for MEM analysis were used,
and the results are shown in Fig. 80.
Over a fairly wide range of frequencies
MEM gives results close to the theoret-
ical line in the nontapered frequency
range, between 0.078-0.156 Hz. The
results of this analysis are encourag-
ing, although the long period part of
the spectrum still presents some prob-
lems. Further experimentation needs
to be done using real data, with various
window lengths.
140
120
100
50
1
WINDOW
i
LENGTH (
sec)
I i
_ ■
O
25
20
* " J^
-
•
1 5
- e
10
e
^^*!
_
„amp =0 | cosine taper i
1 m o
i I
1 cosine taperj amp =
-
i
i £&
#T
-
<" life
i
i
i
1 1
-
0.05
O.IO 0.15
FREQUENCY (HZ)
020
Fig. 80. Maximum entropy frequency analysis of synthetic dispersed wave train using a sliding
window of varying length. Time is taken to be the center of the window. The synthetic signal is of
constant amplitude over the frequency range 0.078-0.156 Hz, and its amplitude spectrum is cosine
tapered to zero over the frequency range shown. Frequencies over which spectral amplitudes are zero
are indicated. The solid sloping line through the points is the theoretical linear dispersion curve.
DEPARTMENT OF TERRESTRIAL MAGNETISM
249
Concluding Remarks
Maximum entropy analysis of syn-
thetic signals of geophysical interest
suggests that this technique may be a
powerful tool. These preliminary in-
vestigations indicate that, at least for
the Brune model and the dispersed
wave, MEM does not suffer from the
signal truncation effects in the same
manner as the Fourier transform. For
the dispersed wave, MEM correctly ob-
tains a linear dispersion curve over a
fairly wide range of frequencies. The
failure to find the curve at the upper
and lower ends of the frequency range
may be due to artificial tapering of the
data. MEM does not seem to provide
the desired results at the low frequency
end of the spectrum of signals such as
the sine wave convolved with instru-
ment response or the double Brune
pulse. It is no t clear why MEM pro-
duces a nonzero plateau at the long-
period end of the amplitude spectrum
yet succeeds on simple sine waves and
a few more complex signals such as the
Brune pulse.
Acknowledgments
This research was supported in part
by National Science Foundation Grant
No. DES 72-01295A02.
References
Akaike, H., Fitting autoregressive models
for prediction, Ann. Inst. Statist. Math.,
21, 243-247, 1969.
Akaike, H., Statistical predictor identifica-
tion, Ann. Inst. Statist. Math., 22, 203-
217, 1970.
Brune, J.N., Tectonic stress and the spectra
of seismic shear waves from earthquakes,
J. Geophys. Res., 75, 4997-5009, 1970,
erratum, 76, 502, 1971.
Burg, J. P., Maximum entropy spectral
analysis, paper presented at the 37th An-
nual International Meeting, Soc. of
Explor. Geophys., Oklahoma City, Okla.,
Oct. 31, 1967.
Burg, J. P., A new analysis technique for
time series data, paper presented at Ad-
vanced Study Institute on Signal Process-
ing, NATO, Enschede, Netherlands,
1968.
Chen, W.Y., and G.R. Stegen, Experiments
with maximum entropy power spectra of
sinusoids, J. Geophys. Res., 79, 3019-
3022, 1974.
Claerbout, J .F ., Fundamentals of Geophysi-
cal Data Processing, McGraw-Hill, New
York, 274 pp., 1976.
Linde, Alan T. and I. Selwyn Sacks, Errors
in the spectral analysis of long-period
seismic body waves, J. Geophys. Res., 76,
3326-3335, 1971.
McGee, T.M., On Berg's method of spectral
analysis, unpublished manuscript, Ven-
ing Meinesz Laboratory, Utrecht, 1969.
Seber, C, Maximum entropy spectral
analysis with particular interest in appli-
cations to seismic waves, Unpublished
LBA Thesis, Kalamazoo College, 1976.
Shannon, C.E. and W. Weaver, The Math-
ematical Theory of Communication, Uni-
versity of Illinois Press, Urbana, 117 pp.,
1949.
Ulrych, Tad J., and Thomas N. Bishop,
Maximum entropy spectral analysis and
autoregressive decomposition, Rev.
Geophys. Space Phys., 13, 183-199, 1975.
Ulrych, T.J., D. E. Smylie, O. G. Jenson,
and G. K. C. Clarke, Predictive filtering
and smoothing of short records by using
maximum entropy, J. Geophys. Res., 78,
183-200, 1973.
ATOMIC PHYSICS AND PROPERTIES OF MATERIALS
L. Brown, A. W. Hofmann, M. Magaritz, and G. H. Pepper
Diffusion in Silicate Melts
and Glasses
Albrecht W. Hofmann and Mordeckai Magaritz
Introduction
Our experimental work on cation dif-
fusion in natural melts was originally
motivated by the need to know more
about the length and time scales of
chemical processes in a partially mol-
ten upper mantle (see Hofmann, Year
Book 73, pp. 935-941, 1974; Year Book
74, pp. 183-189, 1975). For this pur-
pose, calcium and strontium diffusion
were investigated in a natural basalt
250
(olivine tholeiite). We have continued
and expanded this work and report
more complete diffusion data for stron-
tium and data for barium and cobalt in
the same basalt matrix. The results
confirm that at temperatures of
1300°-1400°C, i.e., somewhat above
the basalt liquidus at 1 atm. pressure,
the diffusion coefficients of all divalent
cations so far investigated lie in the
range of 10 7 to 10 _,) cm 2 s" 1 .
In addition, we have started exten-
sive diffusion studies on glass of grani-
tic composition. This glass serves as a
model for melts typical of those occur-
ring in the earth's crust, and the re-
sults are of particular importance to
the interpretation of Rb-Sr age meas-
urements on partially molten crustal
rocks, migmatites. We report results
for Sr and Ba diffusion in this glass. We
find that diffusive transport distances
of strontium in granitic melts do not
exceed a few tens of centimeters in 10"
yr, so that the range of isotopic equilib-
ration of Sr in migmatites is expected
to be small also.
The measured diffusion coefficients
for Ba and Sr in both basaltic and
granitic melts (which have viscosity
differences of several orders of mag-
nitude) demonstrate that the mobility
of these cations is not related in a sim-
ple way to the viscosity of the liquid as
would be predicted by the Stokes-
Einstein relation. Rather, ionic charge
and ionic radius are shown to have a
very strong effect on the mobility, es-
pecially in the lower temperature
range of the obsidian measurements of
675°-890°C. Many ionic species such as
Fe, Co, and Eu can be present in more
than one oxidation state in the silicate
melt, and the contrast in diffusivities
can be used to determine the propor-
tions, for example, of Co +:! /Co +2 in the
melt. The determination of oxidation
state is particularly important for
europium, whose partition coefficient
in plagioclase melt systems depends on
the relative abundance of the 2 + state.
In this way it is advantageous to use
kinetic measurements to determine
equilibrium behavior. Because of its
CARNEGIE INSTITUTION
importance to rare earth geochemistry,
we plan to explore this prospect more
fully in the future.
Diffusion in Basalt Melt
Tracer diffusion of K5 Sr and ' i! Ba was
determined by the thin-source tech-
nique explained in last year's report
(Hofmann, Year Book 74, pp. 183-189,
1975). To overcome the effects of initial
convective movement in the region of
the tracer source, the charges (basalt-
filled Pt tubes) were generally 2.0-2.5
cm long, and run durations were sev-
eral days. The Ba-diffusion and Co-
diffusion measurements were made in
a new high-temperature furnace with a
large (about 5-in diameter) heating
chamber, MoSi 2 heating elements, and
thermocouple (Pt i0 Rh,j) controlled tem-
perature regulation. The sectioning
method was improved by using a bench
micrometer capable of measuring
lengths to about l/xm.
Figure 81 shows the results of the
measurements of the residual activity
of 1:i:i Ba after successive length reduc-
tions by grinding the basalt rod from
the "hot" end. The residual activity for
diffusion from a thin source into a
semiinfinite medium is given by:
A resid /A„ = erfcy
where y = x/2 / Dt
(1)
For convenience of data analysis the
measured activities are converted (us-
ing Equation 1) to the dimensionless
parameter y, which is then plotted as a
function of length removed, x, or as a
function of x/2 /Mvhere t is the dura-
tion of the diffusion experiment. The
analytical data will define a straight
line if the results conform to the initial
and boundary conditions of one-
dimensional Fick's Law diffusion with
constant diffusion coefficient from a
thin source into an infinite medium. A
least-squares regression analysis de-
termines the best line through the
data, and the diffusion coefficient is
given by the square of the slope of the
DEPARTMENT OF TERRESTRIAL MAGNETISM
251
1 1 1 1 1 1 1 I 1 1 1 1 1 1
-
1260 ^V, X*~^I350 x /
>V / J^ /
_
o / / y x ^-1440
-
/ • / / /
-
//// x/
-
1 / £, A
// // x^
-
/• // x
:
- <iy x /
_
/ // / x
o«* /«
r 1 1 1 1 1 1 1 1 1 1 1 1 1 1
-
1.8
1.6
1.4 -
1.2 -
1.0
0.8
0.6 -
0.4 -
0.2 -
5 10 15
I0 4 -x/2nr(cmsec- |/2 )
Fig. 81. Linearized plot of residual activity versus x/2 vTfor m Ba diffusion in olivine tholeiite
basalt: x is the length removed from the cylinder; t, the derivation of the diffusion experiment; and Y
is the dimensionless parameter calculated from the measured residual activity using Equation 1.
Each symbol plotted represents a separate diffusion experiment; the numbers are the temperatures in
degrees Celsius for each experiment.
line. Inspection of Fig. 81 shows that
the data do in general define excellent
straight lines. Significant deviations
occur only in the region near the origin
where convection appears to disturb
the diffusion profiles and where the
melt frequently forms a convex or con-
cave meniscus against air so that the
position for the origin itself is poorly
defined. Occasionally, significant de-
viations are also found for large values
of x/2 JT~ where small errors in the
background, correction cause signifi-
cant systematic deviations from the
straight line.
Figure 82 is an Arrhenius plot (log
diffusion coefficient versus the inverse
absolute temperature) of the new Ba,
Sr, and Co data. Also shown is the Ar-
rhenius line for Ca diffusion previously
reported (Year Book 74, pp. 183-189)
and three additional measurements of
Sr diffusion determined by the
diffusion-couple method reported in
Year Book 73, pp. 935-941. In that re-
port the error for the two lower-
temperature points had been estimated
to be within a factor of 2, and that of
the highest-temperature measurement
had been judged less reliable because of
deformation of the diffusion couple dur-
ing the run. These error estimates ap-
pear to have been realistic; the Ar-
rhenius line is drawn on the basis of
the new thin-source data alone; and the
activation energy of 44 kcal/mol super-
sedes the previously given value of 114
kcal/mol. The extreme error in the
original estimate of the activation en-
ergy is largely a result of the error in-
troduced by the above-mentioned
high-temperature point whose D value
is too large by a factor of 3. The Ar-
rhenius lines given in Fig. 82 are de-
scribed by the equations: D(Ca) =
0.535 exp(-44.0 x WIRT), D(Sr) =
0.278 exp( -43.5 x 1 W/RT), and D(Ba)
= 0.059 exp(-39.4 x 1(P/RT), where R
is the gas constant and T is the abso-
lute temperature.
Figure 82 shows that there is a small
but significant decrease in D values as
a function of increasing ionic radius
(Ca 2+ = 0.99A, Sr 2 * = 1.13A, Ba 2+ =
1.35A Pauling radii). On the other
hand, the activation energies for the
three species (44,44, and 39 kcal/mol)
are not significantly different from one
another. Also shown on Fig. 82 are a
few results for cobalt diffusion. Under
252
CARNEGIE INSTITUTION
1500
1400
T PC)
1300
6.0 6.2 6.4
I0 4 /T(°K _I )
Fig. 82. Log,,, D versus \IT (Arrhenius plot) for l:ii Ba, K5 Sr, 45 Ca, ''"Co diffusion in olivine tholeiite
basalt in air at 1 atm. Triangles = Co, circles = Sr, crosses = Ba. The line representing Ca diffusion is
taken from Hofmann ( 1975).
the conditions of the experiment, dry
heating in air, cobalt ions are expected
to be primarily divalent at tempera-
tures above 1000°C (Coons et al., 1976),
and the diffusivities are very close to
those of calcium. These results conform
roughly to the "compensation law" pos-
tulated by Winchell (1969), which
states that the diffusion coefficient of
all species is the same at some critical
temperature. Winchell and Norman
(1969) found that the critical values
were D ~ lO^cirfs » and T = 1500°C.
In detail, our results differ from this
behavior in that the four Arrhenius
lines are essentially parallel rather
than intersecting at a single point.
Nevertheless, estimates of diffusion
coefficients using the compensation
law appear to be much closer to the
true values than those based on the
Stokes-Einstein relation as done by
Donaldson (1975). In the Stokes-
Einstein relation the diffusion coeffi-
cient is proportional to the inverse of
the viscosity of the liquid. As
Donaldson showed, the estimated dif-
fusion coefficients are within a factor of
10 or 20 of the correct value for basalt
near its liquidus, but this agreement
appears to be fortuitous, because as
will be shown below, the Stokes-
Einstein estimate of the diffusion coef-
ficient fails by many orders of mag-
nitude when it is applied to high-
viscosity silicate liquids such as gran-
ite melts.
When these data are compared with
those given in the companion report by
Hofmann and Brown, it appears that
the diffusion coefficients for many ca-
tions in basalt in the temperature
range of 1300°C to 1400°C are approx-
imately logZ) = -6.5±0.9. If the (un-
known) effects of higher pressure and
possible volatile content can be ig-
nored, the characteristic transport dis-
tance by diffusion in a partially molten
mantle can be estimated using the ex-
pression x — JDt, or about 1 km in 10 9
yr. This very limited chemical trans-
port in the mantle provides some strin-
gent restrictions on the choice of physi-
cally possible models for the chemical
evolution of the mantle. Some of these
have been discussed in previous reports
(Hofmann, Year Book 73, pp. 935-941,
1974; and Year Book 74, pp. 183-189,
1975). The following may serve as an
additional example of the consequences
.of the slow diffusion transport.
Kay (1975) proposed a model for the
upper mantle "involving downward
zone refining within a thickening
lithosphere plate." The lithosphere is
assumed to grow by downward solidifi-
DEPARTMENT OF TERRESTRIAL MAGNETISM
253
cation of a partially molten mantle.
"LIL(= Zarge ion Zithophile) elements
and volatiles are largely excluded from
the crystallizing solid phases and ac-
cumulate in a downward migrating
zone at the base of the plate." The re-
sult is taken to be a zone of LIL
element-enriched mantle that ulti-
mately serves as the source material of
oceanic islands. This model, like other
zone refining processes, requires chem-
ical transport and mixing in the molten
zone. Without such transport the bulk
composition of the solidifying liquid
cannot change and no refining takes
place. In the vicinity of the slowly ad-
vancing solidification front the melt
fraction must be very small. This is re-
quired by the phase relations of ther-
modynamic systems that contain
phases of continuously variable compo-
sition. As the melt fraction decreases
toward the solidification front, the bulk
viscosity of the material increases,
thus inhibiting convection progres-
sively toward the boundary. In addi-
tion, the zone refining mechanism re-
quires that mixing occur on a scale that
is small relative to thickness of the so-
lidified material (so that a significant
enrichment factor can be attained) but
large enough so that the quantity of
enriched material is of some geochemi-
cal significance. Thus, ideally the mix-
ing zone below a 100-km-thick litho-
sphere should be on the order of 10 km
thick. We .must therefore consider a
zone of increased viscosity (relative to
the bulk of the partially molten man-
tle) and restricted size of convection
cells (relative to "normal" mantle con-
vection cells expected to have a charac-
teristic size of 100-1000 km or more).
Simple dimensional considerations
show that the velocity in thermally
driven free convection is proportional
to pagUAT/fjL (Bird et al., 1960, Ch. 9),
where p is the density, /x the viscosity,
a the thermal expansion coefficient, g
the gravitational acceleration poten-
tial, L a characteristic length, and AT"
the temperature difference. Conse-
quently, if we allow a velocity of 10
cm/yr for large plate-driving convec-
tion cells with a minimum cell diame-
ter of 100 km and if the bulk viscosity
in the vicinity (i.e., within 10 km) of
the solidification front increases by a
minimum of a factor of 10, the
maximum velocity in a hypothetical
mixing cell of 10-km diameter would be
on the order of 10~ 2 cm/yr or less — that
is, at least an order of magnitude
slower than the average rate of litho-
sphere growth (assumed to be 100
km/10 8 yr or 0.1 cm/yr). In view of this
it seems reasonable to consider the case
where chemical transport occurs pre-
dominantly by diffusion alone. This
process will produce an LIL element-
enriched zone in the vicinity of the sol-
idification front. If the diffusion coeffi-
cient is 10 _,i cm 2 sec -1 , and the velocity
of the front is 0.1 cm/yr, a steady-state
chemical gradient will evolve within
about 10 5 years, after which the sol-
idifying lithosphere will have a compo-
sition identical to that of the original
partially molten asthenosphere. The
width of the enrichment zone can be
roughly estimated by the relation x =
D/V, where D is the diffusion coeffi-
cient and V is the velocity of the sol-
idification front (see Albarede and Bot-
tinga, 1972). This yields a characteris-
tic distance of 3 meters. It is unlikely
that an enrichment zone of such min-
ute thickness would be of any impor-
tance to the chemical evolution of the
mantle, and Kay's model of downward
zone refining is not expected to be ef-
fective. On the other hand, some degree
of reenrichment of the asthenosphere
as a whole cannot be excluded.
Diffusion in Obsidian Glass
An obsidian sample from Valles
Caldera, New Mexico, was chosen as a
model for granitic melts. Shaw (1963,
1974) had previously used this sample
for measurements of viscosity and
water diffusion. Cylinders were drilled
from slabs of this glass, and radioactive
tracer was loaded on the polished ends
of the preannealed cylinders. Care was
taken to avoid contamination of the
sides of cylinders (1) by loading and
drying only small drops of tracer solu-
254
CARNEGIE INSTITUTION
tion near the center of cylinder face
and (2) by annealing the loaded cylin-
der for about 60 min at 700°C to allow
diffusion into a thin surface layer of the
glass and subsequently dissolving the
excess tracer salt from the surface by
washing in dilute HC1. After this pre-
liminary treatment the cylinders were
annealed at temperatures between
675° and 950°C for periods of several
days. The diffusion profiles were then
determined by the same method of sec-
tioning and residual-activity counting
as previously explained for basalt.
Figure 83A shows two typical activ-
ity profiles for Sr and Ba at 850°C. Fig-
ure 83B shows the same data plotted as
explained above for Fig. 81. Figure 84
is an Arrhenius plot of all the results
for Sr and Ba. Judging from the fit of
the data to the least-squares regression
line, the precision is considerably bet-
ter than was achieved in the basalts.
We ascribe this to the absence of any
convective movement in the case of the
obsidian glass. In addition, the activa-
tion energy (proportional to the slope of
the Arrhenius line) is more precisely
defined because of the greater spread of
measurements along the 1/T axis. The
Arrhenius equations are: D(Sr) = 0.09
exp (-43.8 x WVRT) and D(Ba) =
0.053 exp (-45.8 x 1WRT).
The results are relevant to the in-
terpretation of Rb-Sr isochron meas-
urements of migmatites (rocks of
roughly granitic composition formed by
partial melting of crustal rocks). Mig-
matites are typically inhomogeneous
in composition and appear in layers of
centimeters or tens of centimeters
thickness. Thus it is immediately obvi-
ous that the high-grade metamorphic
process that produced the migmatites
did not cause chemical homogenization
on a scale greater than a few tens of
centimeters. On the other hand,
isotopic homogenization is required
(and frequently assumed to occur in
high-grade metamorphism) in order for
an Rb-Sr isochron to be "reset" by the
metamorphism. The minimum dis-
tance of isotopic equilibration can be
estimated assuming minimum values
for temperature and duration of the
metamorphism. Choosing values of t =
10 (i years for the duration of near-peak
temperature and 700°C for the temper-
ature itself, we obtain D = 10"' 's -1 and
the characteristic transport distance of:
x = fDi = 20 cm.
'0.5
1
1 1
Sr o
I
A
Ba x
-\\
Ge
-
l\\
-
-I \ ,
-1 \
-1
-
1
X
x x
— X
—
i i
200
400
600
x (Mm)
Fig. 83. Diffusion profiles in obsidian for Sr, Ba, and Ge at 850°C. (A) Plot of residual activity
versus length of cylinder removed. The solid lines are ideal diffusion curves, Equation 1.
DEPARTMENT OF TERRESTRIAL MAGNETISM
255
I0 5 ' x/2fl7(cm 2 sec- l/2 )
Fig. 83. (B) Linearized plot of the same data with Y calculated from Equation 1 (see also Fig. 81).
These results show that diffusion of Sr and Ba can be described by a single constant diffusion
coefficient, but that the Ge data either require a variable diffusion coefficient, or have more than one
chemical species of Ge present.
io-
10"
CM
§ io-
10"
900
I
T(°C)
800
I ,
700
-r^ — i 1 r
8
J L
J I I L
10
J I I L
io 4 /t (°k _i )
Fig. 84. Arrtenius plot for Sr (circles) and Ba (crosses) diffusion in obsidian (in air).
256
CARNEGIE INSTITUTION
Such small transport distances must
be considered when samples are taken
for isochron measurements in order to
determine the age of migmatization.
For example, it is common practice to
collect blocks of at least 20 cm diameter
as individual samples and to space
such samplings over meters or even
kilometers. Isochron measurements on
such suites of whole-rock samples are
then frequently interpreted as deter-
minations of the age of metamorphism.
Although nondiffusive migration of
aqueous pore fluids or melt may con-
ceivably have accomplished such
wide-ranging isotopic homogenization
of strontium, that interpretation would
need to be confirmed by small-scale
sampling.
The similarity of inferred distance of
Sr migration and major-element equi-
librium leads to the rule of thumb that
centimeter-scale sampling should be
included for Rb-Sr age determinations
of metamorphic rocks and that the
range of major-element equilibrium in
a rock should be used as a guide to the
presumed range of isotopic equilibrium
needed for the age measurements.
Some caution is in order for the
geological interpretation of these diffu-
sion data because the effect of water on
these diffusivities is unknown. We are
therefore preparing a new set of exper-
iments to be conducted under hy-
drothermal conditions.
The determination of one of the fun-
damental properties of natural melts,
ion mobility, has many applications in
the earth and planetary sciences as
well as in material science. These in-
clude such diverse problems as deter-
mination of quenching times and spa-
tial configurations used in experimen-
tal petrology; interpretation of concen-
tration gradients in natural terrestrial
and lunar glasses; problems of crystal-
lization kinetics, such as the develop-
ment of oscillatory zoning in plagio-
clase (Bottingae^a/., 1966); determina-
tion of the structure of silicate melts;
and measurement of oxidation state of
ionic species in melts. We will discuss
briefly the latter two of these applica-
tions.
It is shown on Fig. 83 that the diffu-
sion profile for Ge in obsidian is not
compatible with a single constant dif-
fusion coefficient. Similar results were
obtained for Co diffusion. Both ele-
ments can be present in more than one
oxidation state, and we have attempted
to deconvolute the Ge profile into two
components, each diffusing indepen-
dently of the other. The solid symbols
shown in Fig. 85 form approximately
straight lines, which correspond to a
diffusion coefficient of D = 1.8 x 10~ 8 .
This was calculated by subtracting
from the measured profiles (open sym-
bols) a slow-moving component, assum-
ing this accounts for 90% of the total
Ge activity and has a diffusion coeffi-
cient of D =3.8 x 10" 12 cm 2 s" 1 . Our
interpretation that the two species in-
volved are two oxidation states of Ge is
subject to confirmation by experiment
under controlled partial pressure of
oxygen. This offers the possibility of
measuring the proportions in which
different oxidation states of the same
chemical element are present in a melt.
To our knowledge, other methods for
determining this are not feasible for
trace elements.
Figure 86 shows a compilation of our
diffusion data. Also shown are tracer
diffusion coefficients for alkalies in ob-
sidian measured by Jambon and Car-
ron (1973) on a single diagram. It is
clear that extrapolation for the obsi-
dian data to the temperature range of
the basalt liquidus yields diffusion
coefficients that differ from the various
basalt diffusivities by factors of about
20 for barium and 2 for strontium. In
contrast, the viscosity of these two
melts varies by about three to four or-
ders of magnitude (Murase and McBir-
ney, 1973). From this it is evident that
the Stokes-Einstein relation (which re-
lates diffusivities to viscosity) is not
applicable to liquids of this kind. Fur-
thermore, the results are compatible
with structure models that consider ca-
tion species such as Ca 2 , Sr 2 ^ and Ba 2+
DEPARTMENT OF TERRESTRIAL MAGNETISM
2.0
257
Y
1.5
1.0
0.5
i 1 1 r
1 r
i r
1 r
1 r
<P
10
15
x/2\IT (cm sec - " 2 )
Fig. 85. Diffusion profiles for Ge in obsidian at 850°C. Open circles and open diamonds denote the
results of the linearization procedure for two separate experiments as used in Figs. 81 and 83; the
solid circles and solid diamonds were calculated by subtracting 90% of the total activity as the
slow-moving Ge component (D = 3.8 x 10~ 12 cm 2 s _1 ) in order to isolate the fast moving component
which yields a straight line Y versus X relationship and a diffusion coefficient of D = 1.8 x
10- 8 cm 2 s-'.
1400 1200 1000
4 i l_, I i , i
T(°C)
800
600
I0 4 /T(°K"')
Fig. 86. Compilation of diffusion data in obsidian (at temperatures less than 1000°C) and basalt
(at temperatures greater than 1250°C). The Arrhenius lines for K, Rb, and Cs are taken from Jambon
and Carron (1973). The points for Ge are derived by using the deconvolution procedure shown in Fig.
85 and assuming that the slow and fast species involved are Ge^ + and Ge 2+ , respectively.
258
CARNEGIE INSTITUTION
as single ions that migrate through the
network of silica tetrahedra at a rate
primarily dependent on ionic size and
charge and nearly independent of the
degree of polymerization of the silica
tetrahedron. Figure 86 shows, for
example, that Ba 2+ and K + , which have
similar ionic radii but different ionic
charge, differ drastically in their mo-
bility in obsidian. At 800°C the tracer
diffusion coefficient for K + is greater by
a factor of 10 ! than the coefficient for
Ba 2+ . In addition, it appears that the
effect of ionic radius differences is more
important for univalent ions than for
divalent ions. Preliminary data for tri-
valent ions (Gd 3+ and Al :i+ , also shown
on Fig. 86) give diffusion coefficients
smaller than about 2 x 10~ 12 cm 2 s _1 at
850°C in obsidian.
The above model for the structure of
silicate liquids contrasts with models
that propose the existence of larger
complexes or "units of structure related
to species or subspecies of minerals"
(Yoder, 1973). These complexes are
thought to mimic the composition of
solid phases that would crystallize
below the liquidus temperature. The
presently available diffusion data are
not compatible with such a model be-
cause the cations are evidently not suf-
ficiently bound to the silica tetrahedra
to form stable complexes.
Acknowledgments
We are grateful to Dr. H. S. Yoder,
Jr., for supplying the basalt sample
and for making the facilities of the
Geophysical Laboratory available to
us. We especially appreciate the use of
a platinum-wound furnace at the
Geophysical Laboratory. We also
thank Dr. R. L. Smith, who kindly
made the sample of obsidian available
for this study. Financial support was
given by the National Science Founda-
tion, Grant No. GA-40250.
References
Albarede, F., and Y. Bottinga, Kinetic dis-
equilibrium in trace element partitioning
between phenocrysts and host lava,
Geochim. Cosmochim. Acta, 36, 141-156,
1972.
Bird, R. B., W. E. Stewart, and E. N. Light-
foot, Transport Phenomena, John Wiley
& Sons, New York, 1960.
Bottinga, Y., A. Kudo, and D. Weill, Some
observations on oscillatory zoning and
crystallization of magmatic plagioclase,
Am. Mineral, 51, 792-806, 1966.
Coons, W. E., J. R. Holloway, and A. Nav-
rotsky, Co 2+ as a chemical analogue for
Fe 2+ in high temperature experiments in
basaltic systems, Earth Planet. Sci. Lett.,
30, 303-308, 1976.
Donaldson, C. H., Calculated diffusion coef-
ficients and the growth rate of olivine in a
basalt magma, Lithos, 8, 163-174, 1975.
Hofmann, A. W., Strontium diffusion in a
basalt melt and implications for Sr
isotope geochemistry and geochronology,
in Carnegie Inst. Wash. Year Book 73, pp.
935-941, 1974.
Hofmann, A. W., Diffusion of Ca and Sr in a
basalt melt, in Carnegie Inst. Wash. Year
Book 74, pp. 183-189, 1975.
Jambon, A., and J. P. Carron, Etude ex-
perimentale de la diffusion des elements
alcalins K, Rb, Cs dans une obsidienne
granitique, C.R. Acad. Sci. D, 276,
3069-3072, 1973.
Kay, R., Chemical zonation of the oceanic
mantle (Abstract), EOS Trans. Am.
Geophys. Union 56, 1077, 1975.
Murase, T., and A. R. McBirney, Properties
of some common igneous rocks and their
melts at high temperatures, Geol. Soc.
Am. Bull., 84, 3563-3592, 1973.
Shaw, H. R., Obsidian-FLO viscosities at
1000 and 2000 bars in the temperature
range 700° to 900°C, J. Geophys. Res., 68,
6337-6343, 1963.
Shaw, H. R. Diffusion of H,0 in granitic
liquids: Part I, Experimental data; Part
II, Mass transfer in magma chambers, in
Geochemical Transport and Kinetics, A.
W. Hofmann, B. J. Giletti, H. S. Yoder,
and R. A. Yund, eds., Carnegie Inst.
Wash.,Publ. 634, pp. 139-170, 1974.
Winchell, P., The compensation law for dif-
fusion in silicates, High Temp. Sci., I,
200-215, 1969.
Winchell, P., and T. H. Norman, A study of
the diffusion of radioactive nuclides in
molten silicates at high temperatures, in
High Temperature Technology, 3rd
Intern. Symp., Asilomar 1967, pp. 479-
492, 1969.
Yoder, H. S., Jr., Contemporaneous basaltic
DEPARTMENT OF TERRESTRIAL MAGNETISM
259
and rhyolitic magmas, Am. Mineral., 58,
153-171, 1973.
Diffusion Measurements Using
Fast Deuterons for in situ
Production of Radioactive
Tracers
A. W. Hofmann, and L. Brown
In connection with our program to
measure diffusion in silicate melts we
have carried out preliminary experi-
ments to determine the feasibility of
creating a sharply defined zone of
radioactive tracers within a sample of
quenched basalt glass. Such a tech-
nique has at least two potential advan-
tages over the more conventional tech-
nique of applying a thin layer of tracer
to the surface of the sample (see com-
panion report by Hofmann and
Magaritz): (1) Relatively volatile ele-
ments such as sodium can be easily
contained within the sample; and (2)
convective movement of the melt dur-
ing initial heating of the sample may
be less severe in the center than at the
surface of the sample.
Sodium is quite volatile at tempera-
tures of 1300° to 1400°C, and much of
an externally applied sodium tracer
would be expected to be lost by surface
evaporation. In addition, the boundary
conditions for the diffusion experi-
ments would be uncertain and temper-
ature dependent. This problem is elim-
inated if the sodium tracer is at the
center of the sample and if the diffusion
experiment is terminated before a sig-
nificant amount of the tracer reaches
the surface.
Convective mixing was found to
cause significant errors in short-term
diffusion experiments (less than one
day) and it was believed that this con-
vection may be confined to a relatively
narrow zone near the open end of the
platinum capillary tube that contains
the melt (see Hofmann, Year Book 74,
pp. 183-189, 1975). If this is correct,
the problem of convective interference
could be circumvented by introducing
the tracer in the center of the Pt tube
and thus avoiding the near-surface
convection. In addition, it is very dif-
ficult to prove conclusively that trans-
port occurred primarily by volume dif-
fusion rather than by some other pro-
cess, perhaps resembling eddy diffu-
sion, unless differential transport be-
tween different chemical species can be
demonstrated.
Experimental Method
The basalt (olivine tholeiite, 1921
eruption of Kilauea) was contained as
quenched glass in 2-mm Pt tubes.
These were exposed to a collimated
d beam of 40 MeV energy from the
Lawrence Berkeley 88-in cyclotron
which had passed through a 0.0025-cm
Al foil and through 2 in of air before
reaching the samples. A slotted 0.178-
cm-thick brass plate was used for col-
limation. The width of the collimation
slit was 1 mm in some experiments and
0.5 mm in others. The exposure varied
between 5 and 40 /i.A-min per sample.
In order to determine the initial distri-
bution of activity within the sample,
one basalt-filled tube was sectioned
after irradiation, and the activity was
measured using a 3 x 3 in Nal well
detector with the sample placed in the
well. Figure 87 shows this initial pro-
file of the 0.847-MeV gamma peak,
which is produced by 5, 'Co within the
basalt. Also shown is the profile for the
0.099-MeV gamma peak combined
with lower-energy Pt-x-ray peaks pro-
duced by I9,i Au and 195 Au within the Pt
tube. The gold isotopes are produced by
deuteron interactions with platinum.
The activity profiles in Fig. 87 show
that the collimation is excellent and
that the activity within the irradiated
zone is uniform. The nonhorizontal
slope on the right side is due to absorp-
tion effects during counting. The small
tail beyond the 1-mm zone amounts to
less than 2% of the total activity and is
considered negligible for the purpose of
these experiments.
A gamma-ray energy spectrum
taken with the Nal detector showed
260
CARNEGIE INSTITUTION
I I
+ 0.847 MeV
o < 0.20 MeV
|~lmm-«)
+-+-
0.2
0.4
0.6
0.
x (cm)
Fig. 87. Profile of residual activity (arbitrary
scale) versus length (x) of cylinder after deuteron
irradiation and before diffusion anneal. A con-
ventional, well-type Nal detector was used for
this profile only. The low-energy profile contains
0.099 MeV gamma and lower-energy Ptr-x rays
and shows the effects of absorption by the sam-
ple: The increase in apparent activity occurs as
absorbing material near the "hot" zone is re-
moved by successive sectioning. The absorption
effect is much less pronounced in the 0.847-MeV
profile for 5H Co. The actual width of the irradia-
tion window is shown by the dashed vertical
lines.
prominent peaks at 0.1, 0.35, 0.51, and
0.85 MeV, and a broadened peak at
about 1.25 MeV. The only energies that
were apparently associated with a
single isotope were due to 196 Au and
5,i Co. Fortunately, Dr. J. Rowe of the
U.S. Geological Survey kindly made a
Ge(Li) detector available for this study.
This permitted us to obtain a high-
resolution spectrum in which the fol-
lowing species were easily resolved:
22 Na, 46 Sc, 48 V, 54 Mn, 56 Co, 57 Co, ,95 Au,
,9,s Au. The only overlap of spectral lines
occurred between the 0.835-MeV and
0.847-MeV peaks of 54 Mn and 56 Co.
Five annealing experiments were
made at a temperature of 1300°C with
run durations varying between 17 min
and 15.5 hr, using the equipment and
technique described in the companion
report by Hofmann and Magaritz (this
Year Book). The results of the 15.5-hr
run are shown in Fig. 88. This clearly
demonstrates the differential transport
between the different chemical species.
Two features argue strongly that this
transport is primarily caused by vol-
ume diffusion rather than some other
process such as convection: (1) the dif-
ference in transport distances between
species, such as Na and V, and (2) the
fit of the profiles to theoretical
volume-diffusion profiles drawn as
solid lines for Na {D = 2.45 x
lO^citfs- 1 ), Co and Mn (D = 4.58 x
10 7 cm 2 s ■) and V (D = 5.03 x
10" 8 cm 2 s~ 1 ). Only Sc does not fit such a
diffusion profile. This may mean that
there are two Sc (chemical) species pre-
sent, perhaps with different oxidation
states and with two different diffusion
coefficients. These results show that
any contribution by convective trans-
port must have been smaller than the
smallest total transport, that of van-
adium. It has been our experience,
however, that a small but significant
convective contribution to the total
transport is not necessarily detected as
a significant deviation from the ideal
diffusion profile. For example, if an in-
finitely thin source were spread out to
a width of 1 mm by convection at the
beginning of the annealing experi-
ment, the resulting profiles after sev-
eral hours of diffusion annealing would
be essentially indistinguishable from
profiles due to pure volume diffusion.
Such convective contributions do, how-
ever, show up quite well when short-
duration and long-duration anneals
are compared (see Hofmann, Year Book
74, pp. 183-189, 1975). We believe
therefore that isothermal time studies
are indispensable in demonstrating the
validity of any experimental technique
concerned with diffusion.
The results of a series of five experi-
ments at 1300°C for Na and Co trans-
port are shown in Fig. 89, where the
apparent diffusion coefficients are plot-
ted against run duration. The total
range of apparent D value is less than a
DEPARTMENT OF TERRESTRIAL MAGNETISM
261
« 0.5
x (cm)
Fig. 88. Diffusion profiles (residual activity versus sample length) for five deuteron-activated
species: 22 Na (1.275 MeV), >' ; Co (0.847 and 1.238 Mev), 54 Mn (0.835 MeV), 4,i Sc (0.889 and 1.121 MeV),
and J8 V (0.983 and 1.312 MeV). The temperature was 1300°C, and the run duration was 15.5 hr. The
solid curves are calculated diffusion profiles for D = 2.45 x 10~ H cm 2 s~'( 22 Na), D = 4.58 x
10- 7 cm 2 s-'( 5,i Coand 54 Mn),and£) = 5.03 X 10 K cm 2 s-'( 48 V).
factor of 3 for Co and less than 1.5 for
Na. In no case do the measured profiles
deviate significantly from calculated
volume diffusion profiles. The results
indicate that some convective trans-
port does occur in some of the experi-
ments. However, the effect is much less
pronounced than that found in a study
of Ca diffusion reported last year
(Hofmann, Year Book 74, pp. 183-189,
1975). In that study a thin source of
45 Ca was applied to the open surface of
the melt at the end of the Pt tube. Also
a Pt furnace with a vertical muffle tube
was used in the Ca study, whereas we
now use a furnace with a completely
enclosed heating chamber, MoSi 2 -
heating elements, and SCR-type tem-
perature control.
We believe that convective move-
ment occurs during heating-up time
(which takes several minutes) and that
its magnitude depends on small dif-
ferences in the heating schedule not
yet under our control. In particular, we
were surprised to find the apparent dif-
fusion coefficients derived from the
shortest (17-min) run to be identical
within analytical error to the values
derived from the longest run (15.5 hr).
This indicates that there was a negli-
gible amount of convective movement
during this short run. This result and
the comparatively small total range of
apparent D values lead us to the esti-
mate that the results obtained from the
longest run (shown in Fig. 88) are cor-
rect within 20% for Na and 50% for Co
and Mn. We are also encouraged by the
excellent agreement obtained for Co
diffusion at 1300°C between the con-
ventional technique using a surficial
thin source (D ((1 = 4.7 x 10~ 7 cm 2 s _1 ; see
companion report by Hofmann and
262
CARNEGIE INSTITUTION
Q
-
] 1
1 1
1
-
o
o
-
"O
o
"
+
_
-
+
+
+
-
g-
-
' 1 1
1 1
1
-
6XICT
t (sec
Fig. 89. Results of isothermal diffusion study for tholeiite basalt at 1300°C under atmospheric
conditions. D is the apparent diffusion coefficient for Na (circles) and Co (crosses), and t is the run
duration.
Magaritz) and the result of this study,
O co = 4.6 x lO^cnr-'s" 1 .
Further improvement in accuracy
should be easily obtainable by simply
increasing the run duration. This was
avoided in the current experiments in
order to prevent the Na profile from
reaching the ends of the cylinders (al-
though this did occur inadvertently in
one experiment for which no D value is
plotted for Na; it also occurred margi-
nally in the experiment shown in Fig.
88). We conclude that this is an excel-
lent technique to study diffusion of cer-
tain chemical species, including the
volatile species Na, and we expect that
it will be applicable also to diffusion
studies at high pressure for which the
more conventional techniques are not
suitable.
The results also show that the mag-
nitude of the diffusion coefficient of ca-
tions is strongly dependent on ionic
charge: Na (D = 2.5 x lO^cmV) is
monovalent, Mn and Co (D = 4.6 x
10" 7 ) are divalent; Sc is generally be-
lieved to be trivalent (although the
shape of the diffusion profile might be
taken to indicate the presence of some
divalent Sc); and V (5.0 x 10" 8 ) is at
least trivalent. This effect is even more
pronounced at lower temperatures (see
the companion report by Hofmann and
Magaritz). We believe that the appar-
ently very simple relationship between
ionic charge and diffusivity will prove
to be an extremely useful tool to de-
termine the oxidation state of transi-
tion elements in natural melts.
Acknowledgments
We are grateful to Dr. H. E. Conzett
and Ms. Ruth Mary Larimer of Law-
rence Berkeley Laboratory for col-
laborating with us and carrying out the
irradiations and to Dr. Jack J. Rowe of
the U.S. Geological Survey in Reston,
Va., for making available a Ge(Li) de-
tector. We are also grateful to the Na-
tional Science Foundation for provid-
ing financial support through Grant
No. GA-40250.
DEPARTMENT OF TERRESTRIAL MAGNETISM
263
Observation of X Radiation
Characteristic of a United Atom
L. Brown and G. H. Pepper
Introduction
During the past year the Van de
Graaff accelerator was used exclu-
sively for the investigation of processes
present in the collision of heavy ions at
energies of a few MeV. Crucial to these
studies, as well as serving as an incen-
tive for them, was a newly purchased
lithium-drifted silicon detector capable
of resolving 800-eV spectral lines. This
instrument, which is decidedly
superior to the one we previously used
through loan, has allowed us to
examine regions of the x-ray spectrum
that have disclosed two phenomena:
One is the clearest manifestation yet
observed of the formation of a united
atom; the other is a selective transfer of
an electron vacancy from a K shell to a
single L subshell, a form of transfer not
heretofore recognized.
When considering the collision of
heavy ions wherein the electron shells
of the two atoms interpenetrate, it is in
some cases useful to consider them as
having formed a united atom, i.e., an
electronic system that has the nuclei of
both target and projectile briefly form-
ing a common center of attraction. This
view, with its attendant idea of the
formation of states in the united atom
by way of the transformation from
separated-atom levels through
quasimolecular states, has difficulties,
some of which were treated in last
year's Annual Report, but it has cer-
tain elements of truth. When heavy
ions collide, a continuum of x rays is
observed (Macdonald et al., 1973) that
can result from neither nucleus-
nucleus nor electron-nucleus brems-
strahlung. This continuum is generally
interpreted as resulting from a transi-
tion of an electron to a vacancy in a
molecular state formed during the col-
lision. Since such states have energies
that vary with time, these transitions
obviously form continua. However,
transitions to orbitals terminating on a
given united atom shell should have a
limit given by the binding energy of
that shell. This has not been observed
for the K band. Instead, an
exponential-shaped continuum is seen
to extend well beyond the united atom
limits.
The energies attainable with our ac-
celerator do not allow an excitation of
the K band sufficiently intense for
study but do allow excitation of the L
and M bands. These have been ob-
served by others (Mokler et al., 1975)
and are more suggestive of a united
atom than the K band, but these previ-
ous observations suffer from the inter-
ference of characteristic lines with the
continuum and from few data showing
the dependence of the continua on
atomic number and projectile velocity.
We have found a projectile-target com-
bination that allows these two de-
ficiencies to be removed and that shows
continua, both of L and M bands, with
easily recognized limits, which move
exactly as predicted when the atomic
number of the target is changed. Ob-
servations of the projectile velocity de-
pendence of the continua show two ob-
vious features: (1) Near the united-
atom limit the cross section increases
rapidly near the threshold and then
levels off at high energies; (2) the con-
tinuum extends beyond the united-
atom limit and its width increases with
projectile velocity.
The lifetimes of united atoms could
be estimated by dividing the radius of
the shell in question by the projectile
velocity. These lifetimes then give an
upper limit to the uncertainties in the
energies of the states through simple
application of the Heisenberg uncer-
tainty principle. It has been clear from
the beginning that the observed exten-
sion of the spectrum beyond the
united-atom limit follows from this,
but previous attempts to explain
K-band spectra theoretically (Macek
and Briggs, 1974; Miiller, 1976) have
been in vain. Two circumstances com-
bine to make interpretation difficult:
264
CARNEGIE INSTITUTION
(1) No united- atom limit has been ob-
served for the K band; (2) the observed
broadening increases as the square
root of the projectile velocity, whereas
a simple approach to Heisenberg
broadening would seem to require a
broadening proportional to the veloc-
ity.
We observe a clear limit for L and M
bands, as shown in Figs. 90 and 91. Its
dependence on atomic number is
Photon Energy
Fig. 90. The Z dependence of united-atom
limits. The top three spectra show continua re-
sulting from vacancies in the target L shell (2p
and 2s) that transform through molecular orbi-
tals into united-atom L shells; transformations
are shown morphologically at the upper right.
The energy of the 2s united-atom subshell is
shown by vertical arrows for sodium (Z = 11)
projectiles onto targets of titanium, iron, and zinc
(Z = 22, 26, and 30, respectively). The spectra are
normalized at the united-atom limit. The lower
two spectra show the same phenomenon for
M-shell limits produced by sodium projectiles on
targets of palladium and tin (Z = 46 and 50) but
use a different manner of presentation. The pal-
ladium spectrum has been shifted 285 eV to the
right, the difference in the 3s energies of the
united atoms Z = 57 and 61. This shows the iden-
tical shape of the two continua beyond the
united-atom limit. This manner of presentation
could have been used with equal effect for the
L-limit data.
'10-2
10-2-
Photon Energy (keV)
Fig. 91. The velocity dependence of the
broadening beyond the united-atom limit. The
continuum extends beyond the united-atom limit
because that energy must be uncertain by an
amount AE = fy/At, where At is the lifetime of the
state, a lifetime dependent on the projectile ve-
locity. Spectra for three projectile energies are
shown with smooth curves, which are halves of
Gaussians that have been corrected for thick
target effects, fitted to the portion beyond the
united-atom limit of Na + Ni (11 + 28). The
smooth curves are reproduced at the top to bring
out the velocity dependence of the data. If the
effects of detector resolution are removed, the
broadening is found to be proportional to the
square root of the projectile velocity.
exactly as predicted. The broadening
beyond the united-atom limit is very
nearly half of a Gaussian, the flank of
which drops more than two orders of
magnitude; this appearance of the
united-atom limit was expected when
the molecular-orbital continuum was
discovered, but determined searches
failed to reveal anything but the expo-
nentially shaped continuum. We
undertook a study of the energy de-
pendence of the broadening and found
that it was proportional to the square
root of the projectile velocity. Although
in agreement with the K-band studies,
this was somewhat disappointing be-
DEPARTMENT OF TERRESTRIAL MAGNETISM
265
cause we had hoped that this simple
manifestation would have the equally
simple velocity dependence, i.e., a
broadening proportional to velocity. A
curious aspect of the matter is that the
only theories now under serious con-
sideration predict a square-root de-
pendence but do so on the basis of as-
sumptions that seem to us questiona-
ble. Since our data are the best verifi-
cation of this dependence, our pre-
judices to the contrary notwithstand-
ing, we are seeking to remove the weak
elements of these theories or to find an
acceptable alternate set of assump-
tions.
Vacancy Sharing in Heavy-Ion
Collisions
The most striking phenomenon of
heavy-ion collisions is the large proba-
bility for producing electron holes in
the less tightly bound of two inter-
penetrating shells that have nearly
equal binding energies, e.g., K shells of
two atoms with Z, ~ Z 2 . Conventional
theory predicts a small probability, if
applied strictly. The novel approach
proposed here last year has found little
acceptance because it predicts only the
relative Z dependence (quite satisfac-
torily) and utilizes a mechanism not
observed elsewhere. Whatever the de-
tails of its operation, molecular excita-
tion depends in some way on the trans-
formation of atomic into molecular
wave functions, and its complete eluci-
dation is the central problem of this
field of research.
If one sets aside the mysterious
manner in which vacancies appear in
the more weakly bound shell and con-
siders instead the method by which
vacancies appear in the more strongly
bound shell, then one finds things in a
more satisfactory condition. Meyerhof
(1973) demonstrated that, given the
vacancy in a light K shell, he could cal-
culate the probability of its transfer to
the heavy K shell. The mechanism for
this is radial coupling, which links
molecular states of the same parity.
We set about to see if this process
also coupled K and L shells and noted
an unusual circumstance best illus-
trated with the raw data. In Fig. 92a is
shown the spectrum of praseodymium
produced by bombardment with 2-MeV
Li + ; it is a normal example of Coulomb
excitation, showing all the L lines that
the detector is capable of resolving
with intensities easily understood
through conventional theory. In Fig.
92b is shown the spectrum of
praseodymium when bombarded with
2-MeV K + ; the K lines of potassium are
the most prominent feature and result
from the molecular excitation dis-
cussed previously. L lines from
praseodymium are also shown but are
in marked contrast to the spectrum of
Fig. 92a, as the total intensity is much
less and the intensity ratio is entirely
different; indeed only the L a and L^ 2
lines are observed.
Reference to x-ray tables quickly dis-
closes that these lines are transitions
to vacancies in the 2p 3/2 subshell and
that transitions to the other L sub-
shells, the 2p 1/2 and 2s 1/2 , are absent
within the observational limits im-
posed by the continuum radiation. The
transfer of vacancies from the K shell
of potassium, if the process is a trans-
fer, is obviously selective. A study of
the velocity dependence of the ratio of
potassium K lines to praseodymium L
lines shows values that lie along an ex-
ponential in 1/v, which present in-
terpretations would indicate as charge
exchange.
Our tentative interpretation of the
process can be understood by reference
to Fig. 93a, which shows the manner by
which the atomic energy levels trans-
form through molecular orbitals into
the energy levels of a united atom for
the collision of K + Pr. One notices
that the potassium K shell transforms
by a 3dcr orbital which has the initial
vacancy and the three praseodymium L
subshells by a 2p77, 2pcr, and 2scr. Ra-
dial coupling proceeds between states
of the same molecular parity, here be-
tween the 3dcr and the two orbitals 2pcr
and 2scr. Rotational coupling between
266
■2 10
-6
I0" 8 -
i r~i i i i — i — r
Pr-L
a
•03.
.CvJ
1 t~)
CVJ
K
/
I I I
I V I
CARNEGIE INSTITUTION
1 I I 1 1 1
K-K
o
Pr-L
a
'■■ CM
01
5 5
E Ph (keV)
Fig. 92. The spectra of praseodymium L lines produced by (a) direct excitation with 2-MeV Li + and
(b) vacancy sharing with 2-MeV K + . The identity of various praseodymium L lines capable of being
resolved by the detector are indicated for both spectra with the location of the potassium K lines also
shown for spectrum (b). Unmarked peaks are the result of target contaminants and a photon escape
peak, an instrumental effect. The L lines of praseodymium in spectrum (b) are thought to result
primarily from the transfer of vacancies from the potassium K shell to the praseodymium L III
subshell (2p 3/2 ) through rotational coupling.
-Ne + Ne
Fig. 93. Correlation diagrams for the collision of (a) K + Pr and (6) Ne + Ne. In both diagrams the
morphology of the transformation of separated-atom through molecular levels to united-atom energy
states is shown. The vertical axis is proportional to the logarithm of the energy; the horizontal axis
shows the internuclear distance schematically. The energy scales for (a) and (b) are different.
DEPARTMENT OF TERRESTRIAL MAGNETISM
267
orbitals of opposite parity is also
known to occur, but it has been ob-
served only in collisions in which the
exchanging orbitals attained nearly
the same energy at small radii of sep-
aration, an example of which is shown
in Fig. 93b. This shows the correlation
of orbitals in the collision of neon and
neon, a collision in which the transfer
of a vacancy from the 2p shell of an
incident ion is transferred by the 2pcr
to the 2p7r orbital. The theory of this
transfer (Fastrup, 1976) agrees well
with experiment. Except at very low
velocities, the probability of transfer is
nearly unity because the two orbitals
have nearly the same energy at small
values of internuclear radius.
Because of the different probability
observed for 3dcr — » 2p7r compared with
3dcr — > 2po- and 2sct, we think that it
must be a new manifestation of rota-
tional coupling — a case of rotational
coupling at large rather than small
internuclear radius.
References
Fastrup, B., Experimental studies of
inner-shell excitation in slow ion-atom
collisions, in The Physics of Electronic
and Atomic Collisions, Invited Papers of
the IX ICPEAC, p. 361, J. S. Risley and R.
Geballe, ed., University of Washington
Press, Seattle, 1976.
Macdonald, J. R., M. D. Brown, and T.
Chiao, Observation of a K x-ray band
emitted by the transient C-C system
formed at keV energies, Phys. Rev. Lett.,
30, All, 1973.
Macek, J. H., and J. S. Briggs, Collision
broadening and molecular orbital x-rays
II, J. Phys., B7, 1312, 1974.
Meyerhof, W. E., K- vacancy sharing in
near-asymmetric heavy-ion collisions,
Phys. Rev. Lett., 31, 1341, 1973.
Mokler, P. H., S. Hagmann, P. Armbruster,
G. Kraft, H. J. Stein, K. Rashid, and B.
Fricke, Superheavy quasimolecules, in
Atomic Physics, 4, G. zu Putlitz, E. W.
Weber, and A. Winnacher, eds., Plenum
Press, New York, 1975.
Miiller, B., Radiative processes in transient
quasimolecules, in The Physics of Elec-
tronic and Atomic Collisions, Invited Pa-
pers of the IX ICPEAC, p. 481, J. S. Risley
and R. Geballe, eds., University of Wash-
ington Press, Seattle, 1976.
268
CARNEGIE INSTITUTION
BIBLIOGRAPHY
Publications listed below can be obtained at no
charge from the Department of Terrestrial Mag-
netism, 5241 Broad Branch Rd., N.W., Washing-
ton, D.C. 20015. When ordering, please give re-
print number(s). No reprints are available for ti-
tles marked with asterisks.
Aldrich, L. T., see Ocola, L. C.
Arculus, R. J., see Shimizu, N.
Assousa, G. E., see Brown, L., and
Warner, J. W.
Boulter, D., see Wallace, D. G.
Brooks, C, see James, D. E.
4415 Brown, L., G. E. Assousa, N. Thonnard,
H. A. Van Rinsvelt, and C. K.
Kumar, Characteristic and con-
tinuum x rays produced with potas-
sium ions of a few MeV, Phys. Rev.,
A12, 425-433, 1975.
Brown, L., see Pepper, G. H., and
Rohrer, U.
Cuyubamba, A., see James, D. E.
Duggan, J. L., see McDaniel, F. D.
Ford, W. K., Jr., see Rubin, V. C.
Gardner, R. K., see McDaniel, F. D.
Gettrust, J. F., see Ocola, L. C.
Gray, T. J., see McDaniel, F. D.
*4416 Hart, S. R., Basement rock synthesis:
Geochemistry, petrology, physical
properties, and paleomagnetism, in
Initial Reports of the Deep Sea Dril-
ling Project, Vol. 34, 814 pp., T. L.
Vallier, ed., Washington, D.C, U.S.
Government Printing Office, 1976.
*4417 Hart, S. R., LIL-element geochemistry,
Leg 34 basalts, in Initial Reports of
the Deep Sea Drilling Project, Vol.
34, 814 pp., T. L. Vallier, ed., Wash-
ington, D.C, U.S. Government
Printing Office, 1976.
Hart, S. R., see Yeats, R. S.
*4418 James, D. E., C. Brooks, and A.
Cuyubamba, Andean Cenozoic vol-
canism: Magma genesis in the light
of strontium isotopic composition
and trace-element geochemistry,
Geol. Soc. Am. Bull., 87, 592-600,
1976.
Kumar, C. K., see Brown, L.
Kushiro, I., see Shimizu, N.
Lear, R. D., see McDaniel, F. D.
Light, G. M., see McDaniel, F. D.
*4419 McDaniel, F. D., T. J. Gray, R. K.
Gardner, G. M. Light, J. L. Duggan
H. A. Van Rinsvelt, R. D. Lear, G. H
Pepper, J. W. Nelson, and A. R
Zander, K-shell x-ray production
cross sections of selected elements Ti
to Sb for 1.0- to 5.0-MeV/amu 7 Li
ions, Phys., Rev., A12, 1271-1280
1975.
Meyer, R. P., see Ocola, L. C.
4420 Mirabel, I. F., and K. C. Turner, An
anomalous velocity neutral hydro-
4421
4422
4423
*4424
4425
4426
4427
4428
4429
gen structure near the galactic cen-
ter, Astrophys. Space Sci., 38, 381-
394, 1975.
Nelson, J. W., see McDaniel, F. D.
Ocola, L. C, L. T. Aldrich, J. F. Get-
trust, R. P. Meyer, and J. E.
Ramirez, S. J., Project Narino I:
Crustal structure under southern
Colombia-northern Ecuador Andes
from seismic refraction data, Bull.
Seis. Soc. Am., 65, 1681-1695, 1975.
Pepper, G. H., and L. Brown, The
15 N(p,a) 12 C reaction with polarized
protons from 0.34 to 1.21 MeW,Nucl.
Phys.,A260, 163-171, 1976.
Pepper, G. H., see McDaniel, F. D.
Peterson, C. J., see Rubin, V. C.
Ramirez, J. E., S. J., see Ocola, L. C.
Rohrer, U., and L. Brown, The
9 Be(p,n) 9 B reaction with polarized
protons from 2.4 to 2.9 MeV, Nucl.
Phys.,A261, 141-148, 1976.
Rubin, V. C, W. K. Ford, Jr., and C J.
Peterson, Evidence for contraction
in the nuclear ring of the barred spi-
ral galaxy NGC 3351, Astrophys. J.,
199, 39-48, 1975.
Rubin, V. C, N. Thonnard, and W. K.
Ford, Jr., Observations of NGC
6764, a barred Seyfert galaxy, As-
trophys. J., 199, 31-38, 1975.
Rubin, V. C, see Warner, J. W.
Shimizu, N., Geochemistry of ul-
tramafic inclusions from Salt Lake
Crater, Hawaii, and from Southern
African kimberlites, in Physics and
Chemistry of the Earth, Vol. 9, pp.
655-669, L. H. Ahrens, J. B. Daw-
son, A. R. Duncan, and A. J. Erlank,
eds., New York, Pergamon Press,
1975.
Shimizu, N., and R. J. Arculus, Rare
earth element concentrations in a
suite of basanitoids and alkali
olivine basalts from Grenada, Les-
ser Antilles, Contrib. Mineral. Pet-
rol, 50, 231-240,1975.
Shimizu, N., and I. Kushiro, The parti-
tioning of rare earth elements be-
tween garnet and liquid at high
pressures: Preliminary experi-
ments, Geophys. Res. Lett., 2, 413-
416,1975.
Snoke, J. A., Archambeau's elas-
todynamical source-model solution
and low-frequency spectral peaks in
the far-field displacement
amplitude, Geophys. J. R. Astron.
Soc, 44, 27-44, 1976.
Thonnard, N., see Brown, L., and Ru-
bin, V. C
Turner, K. G, see Mirabel, I. F.
Van Rinsvelt, H. A., see Brown, L., and
McDaniel, F. D.
DEPARTMENT OF TERRESTRIAL MAGNETISM
269
*4430 Wallace, D. G., Prediction of the sec-
ondary and tertiary structure of
plastocyanin, Biophys. Chem., 4,
123-130, 1976.
*4431 Wallace, D. G., The sequence-
immunology correlation among
higher plant plastocyanins, Im-
munochemistry , in press.
*4432 Wallace, D. G., and D. Boulter, Im-
munological comparisons of higher
plant plastocyanins, Phytochemis-
try, 15, 137-141, 1976.
4433 Warner, J. W., G. E. Assousa, and
others, Radio spectra of OH471 and
OQ112, Nature, 249, 743-746, 1974.
4434 Warner, J. W., and V. C. Rubin, Physi-
cal conditions and structure in NGC
7293, "The Helix," Astrophys. J.,
198, 593-603, 1975.
4435 Wetherill, G. W., Late heavy bom-
bardment of the moon and terrest-
rial planets, Proc. 6th Lunar Sci.
Conf.,2, 1539-1561, 1975.
4436 Wetherill, G. W., Radiometric chronol-
ogy of the early solar system, in An-
nual Review of Nuclear Science, Vol.
25, pp. 283-328, E. Segre, J. R.
Grover, and H. P. Noyes, eds., An-
nual Reviews, Inc., Palo Alto,
California, 1975.
*4437 Yeats, R. S., and S. R. Hart, Introduc-
tion and principal results, Leg 34,
Deep Sea Drilling Project, in Initial
Reports of the Deep Sea Drilling
Project, Vol. 34, 814 pp., T. L. Val-
lier, ed., Washington, D.C., U.S.
Government Printing Office, 1976.
Zander, A. R., see McDaniel, F. D.
Abstracts and Minor
Publications
Ballard, G. S.,see Evans, J. R.
Beavan, R. J., see Evans, J. R., and King, G. C. P.
Bilham, R. G.,see Evans, J. R.
Brooks, C, D. James, and S. R. Hart, Mantle
isochrons and their implications for the
isotopic composition of subcontinental litho-
sphere (abstr.), EOS Trans. Am. Geophys.
Union, 57, 351, 1976.
Evans, J. R., G. S. Ballard, R. J. Beavan, R. G.
Bilham, G. John, and G. C. P. King, A tidal
strain survey of Great Britain (abstr.), EOS
Trans. Am. Geophys. Union, 57, 233, 1976.
Evans, J. R., see King, G. C. P.
Ford, W. K., Jr., see Peterson, C. J., and Rubin,
V.C.
Graham, J. A., see Rubin, V.C.
Hart, S. R., see Brooks, C.
James, D., see Brooks, C.
John, G., see Evans, J. R.
King, G. C. P., I. S. Sacks, J. R. Evans, R. J.
Beavan, and A. Snoke, Single site phase veloc-
ity measurement (abstr.), EOS Trans. Am.
Geophys. Union, 57, 286, 1976.
King, G. C. P., see Evans, J. R.
Linde, A. T., I. S. Sacks, and J. A. Snoke, A nu-
merical model study of multiple rupture
earthquakes (abstr.), EOS Trans. Am.
Geophys. Union, 57, 154, 1976.
Linde, A. T., see Sacks, I. S.
Nicholls, I. A., see Whitford, D. J.
Pepper, G. H. and L. Brown, An observation of
the united-atom limit for the molecular orbi-
tal continuum (abstr.), Bull. Am. Phys. Soc,
21, 649, 1976.
Peterson, C. J., Rotation of the nuclear region of
M31 (abstr.), Bull. Am. Astron. Soc, 7, 254,
1975.
Peterson, C. J., V. C. Rubin, and W. K. Ford, Jr.,
The peculiar galaxy NGC 1275 (abstr.), Bull.
Am. Astron. Soc, 8, 297, 1976.
Peterson, C. J., V. C. Rubin, W. K. Ford, Jr., and
N. Thonnard, Motions of the stars and excited
gas in the barred spiral galaxy NGC 3351
(abstr.), Bull. Am. Astron. Soc, 7, 538, 1975.
Peterson, C. J., see Rubin, V. C.
Roberts, M. S., see Rubin, V. C.
Rodriguez B., A., see Sacks, I. S.
Rubin, V. C, and W. K. Ford, Jr., The motion of
our galaxy and the local group of galaxies
(abstr.), Bull. Am. Astron. Soc, 7, 253, 1975.
Rubin, V. C, C. J. Peterson, and W. K. Ford, Jr.,
The rotation curve of the E7/S0 galaxy NGC
3115 (abstr.), Bull. Am. Astron. Soc, 8, 297,
1976.
Rubin, V. C, N. Thonnard, W. K. Ford, Jr., J. A.
Graham, and M. S. Roberts, A galaxy of coeffi-
cients: A,„ A-,, A„,A LC ,D,„ andLX ( abstr. ),Bull.
Am. Astron. Soc, 7, 531, 1975.
Rubin, V. C, see Peterson, C. J.
Sacks, I. S., and A. Snoke, Heterogeneous veloc-
ity structure at the base of the mantle (abstr.),
Trans. Am. Geophys. Union, 57, 284, 1976.
Sacks, I. S., J. A. Snoke, A. T. Linde, and A. Rod-
riguez B., Shallow seismicity in subduction
zones (abstr.), 47th Annual Meeting of the
Eastern Section of the Seismological Society of
America held in St. Louis, Missouri,
November 6-7, 1975.
Sacks, I. S., see King, G. C. P., Linde, A. T., and
Snoke, J. A.
Snoke, J. A., Implications of moving boundaries
in elastodynamics: Comments on "the pulse
shapes and spectra of elastic waves generated
when a cavity expands in an initial shear
field" by Robert Burridge, J. Geophys. Res.,
81, 1035-1036, 1976.
Snoke, J. A., and I. S. Sacks, Determination of
the lithosphere-asthenosphere boundary
using converted waves (abstr.), EOS Trans.
Am. Geophys. Union, 57, 283, 1976.
Snoke, J. A., see King, G. C. P., Linde, A. T., and
Sacks, I. S.
Thonnard, N., The 21-cm structure of WLM and
its surroundings (abstr.), Bull. Am. Astron.
Soc, 7, 550, 1975.
Thonnard, N., see Peterson, C. J., and Rubin, V.
270 CARNEGIE INSTITUTION
Wetherill, G. W., Total number of earth-crossing Whitford, D. J., and I. A. Nicholls, Potassium
Apollo objects with radii larger than 500 me- variation in quaternary lavas across the
ters (abstr.), EOS Trans. Am. Geophys. Union, Sunda arc in Java and Bali (abstr.), EOS,
57, 275, 1976. Trans. Am. Geophys. Union, 57, 347, 1976.
PERSONNEL
Staff Members
L. Thomas Aldrich David E. James
George E. Assousa Alan T. Linde
Louis Brown Nancy R. Rice 4
Dean B. Cowie 1 Richard B. Roberts 5
Scott E. Forbush, Emeritus Vera C. Rubin
W. Kent Ford, Jr. I. Selwyn Sacks
Stanley R. Hart 2 Norbert Thonnard
Albrecht W. Hofmann Kenneth C. Turner
Bill H. Hoyer :) George W. Wetherill, Director
Distinguished Service Member of Carnegie Institution
Merle A. Tuve
Research Associates (Staff)
Tom I. Bonner 11 J. Arthur Snoke
Mordeckai Magaritz 7 Kiyoshi Suyehiro 8
Research Associates (Nonresident)
Mateo Casaverde, Instituto Geofisco del Hiromu Okada, Hokkaido University, Sap-
Peru, Lima, Peru poro, Japan 9
Leonidas Ocola, Instituto Geofisico del Anibal Rodriguez B., Universidad Nacional
Peru, Lima, Peru de San Agustin, Arequipa, Peru
Shigeji Suyehiro, Japan Meteorological
Agency, Tokyo, Japan 10
Carnegie Fellows (Postdoctoral)
John R. Evans" R. Sundar Rajan 12
George H. Pepper Stuart J. Weidenschilling 1 '
Charles J. Peterson David J. Whitford 14
■Institut Pasteur, Paris, France, from Jan. 20, K From April 1, 1976.
1975. 9 Through Dec. 31, 1975.
^Resigned Aug. 31, 1975. 10 Through Dec. 31, 1975.
'Resigned June 30, 1975. "From Jan. 1, 1976.
'Resigned Feb. 15, 1976. l2 From Sept. 24, 1975.
"'Retired Dec. 31, 1975. ' 'From Feb. 1, 1976.
"Resigned Feb. 15, 1976. 14 From Dec. 1, 1975.
7 FromFeb. 1,1976.
DEPARTMENT OF TERRESTRIAL MAGNETISM
Carnegie Fellows (Predoctoral)
271
John R. Bannister, University of Alberta, Arturo Cuyubamba, McGill University,
Edmonton, Canada Montreal, Canada 15
Visiting Investigators
Claude Allegre, University of Paris, Paris, Montreal, Montreal, Canada 17
France 1 " John R. Cleary, Australian National Uni-
Christopher Brooks, Universite de versity, Canberra, Australia 18
Charles L. Bennett 1 a
Gregory S. DeWitt 2 "
Trainee Fellows
Antonio Flores 21
Michael Schneider-
Christine M. Seber 2 '
Collaborators
E
E
C. Aguirre B., Universidad Mayor de San
Andres, La Paz, Bolivia
Arnal, Institute Argentino de Radioas-
tronomia, Villa Elisa, Argentina
G. Arnold, Ohio State University, Co-
lumbus, Ohio
Bajaja, Institute Argentino de Radioas-
tronomia, Villa Elisa, Argentina
B. Balick, University of Washington, Seat-
tle, Washington
R. J. Beavan, Lamont Doherty Geological
Observatory, Palisades, New York
E. Berg, University of Hawaii, Honolulu,
Hawaii
K. Bloomfield, Institute of Geological Sci-
ences, London, England
M. Bracamonte, Trujillo, Peru
B. F. Burke, Massachusetts Institute of
Technology, Cambridge, Massachusetts
R. Cabre, S. J., Observatorio San Calixto,
La Paz, Bolivia
G. N. Cohen, Institut Pasteur, Paris,
France
R. Colomb, Instituto Argentino de Radioas-
tronomia, Villa Elisa, Argentina
H. E. Conzett, Lawrence Berkeley Labora-
tory, Berkeley, California
G. L. Davis, Geophysical Laboratory, Car-
negie Institution of Washington, Wash-
ington, D. C.
S. del Pozo, Instituto Geofisico Boliviano,
La Paz, Bolivia
J. W. Erkes, State University of New York,
Albany, New York
A. J. Erlank, University of Capetown, Ron-
debosch, South Africa
D. W. Evertson, University of Texas at
Austin, Texas
E. Filloy, Instituto Argentino de Radioas-
tronomia, Villa Elisa, Argentina
C. Garavito, Planetario Destrital, Bogota,
Colombia
L. Gelinas, Genie Mineral, Ecole Polytech-
nique, Montreal, Quebec, Canada
A. A. Giesecke, Instituto Geofisico del Peru,
Lima, Peru
M. Gil, Instituto Argentino de Radioas-
tronomia, Villa Elisa, Argentina
Nicol Girardin, Institut de Physique du
Globe, Paris, France
M. Gordon, Instituto Argentino de Radioas-
tronomia, Villa Elisa, Argentina
J. A. Graham, Cerro Tololo Interamerican
Observatory, La Serena, Chile
C. Heiles, University of California, Berke-
ley, California
C. E. Helsley, University of Texas, Dallas,
Texas
D. Huaco, Instituto Geofisico del Peru,
Lima, Peru
'•'Through Dec. 31,1975.
"'From June 15, 1976.
''Through Dec. 31, 1975.
IK From Sept. 15 to Dec. 31, 1975.
'"May 15- Aug. 15, 1975, and June 1-July 31,
1976.
2 "From June 1, 1976.
-'From May 1 to Aug. 31, 1975.
22 From Sept. 29, 1975 to Mar. 3, 1976.
2! From Jan. 1 to Mar. 31, 1976.
272
CARNEGIE INSTITUTION
G. C. P. King, University of Cambridge,
Cambridge, England
C. K. Kumar, Howard University, Wash-
ington, D. C.
I. Kushiro, Geophysical Laboratory, Car-
negie Institution of Washington, Wash-
ington, D. C.
Ruth M. Larimer, Lawrence Berkeley Lab-
oratory, Berkeley, California
R. P. Meyer, University of Wisconsin,
Madison, Wisconsin
I. Mirabel, Instituto Argentino de Radioas-
tronomia, Villa Elisa, Argentina
R. Morras, Instituto Argentino de Radioas-
tronomia, Villa Elisa, Argentina
Y. Motoya, Kamikineusu Seismological
Observatory of Sapporo University,
Hokkaido, Japan
G. Olafsson, Akureyri, Iceland
J. Oort, Sterrewacht, Leiden, The Nether-
lands
R. W. Page, Bureau of Mineral Resources,
Canberra, Australia
E. Pimental, Cuzco University, Cuzco, Peru
W. Poppel, Instituto Argentino de Radioas-
tronomia, Villa Elisa, Argentina
G. Poupeau, Centre des Faibles Radioacti-
vites, Gif-sur-Yvette, France
J. E. Ramirez, S. J., Instituto Geofisico de
los Andes Colombianos, Bogota, Colom-
bia
M. S. Roberts, National Radio Astronomy
Observatory, Charlottesville, Virginia
R. Rodriguez, Observatorio San Calixto, La
Paz, Boliva
J. Rowe, U.S. Geological Survey, Reston,
Virginia
R. H. Sanders, National Radio Astronomy
Observatory, Charlottesville, Virginia
J. G. Schilling, Narragansett Marine Labo-
ratory, University of Rhode Island,
Kingston, Rhode Island
D. W. Schwartzman, Howard University,
Washington, D. C.
R. G. Seyler, Ohio State University, Co-
lumbus, Ohio
H. Sigtrygsson, Reykjavik, Iceland
D. Simoni, Arequipa, Peru
R. Stefansson, Reykjavik, Iceland
L. Tamayo, Universidad Nacional de San
Agustin, Arequipa, Peru
J. L. Telleria, Planetario Destrital, Bogota,
Colombia
P. Truffa-Bachi, Institut Pasteur, Paris,
France
F. Volponi, Universidad Nacional de Cuyo,
San Juan, Argentina
R. E. White, Steward Observatory, Univer-
sity of Arizona, Tucson, Arizona
J. G. Williams, Jet Propulsion Laboratory,
Pasadena, California
Y. Yamagishi, Matsushiro Seismological
Observatory of Japan, Meteorological
Agency, Matsushiro, Japan
E. Zinner, McDonnel Center for Space Sci-
ences, Washington University, St. Louis,
Missouri
Supporting Staff
Caroline A. Busch, Typist, Fiscal
Assistant 24
R. Paul Busch, Caretaker (temporary)
Louis S. Carter, Instrument Maker
Gloria J. Counts, Clerk-Typist- 5
Dorothy B. Dillin, Librarian
John B. Doak, Electronics Research
Specialist
William N. Dove, Office Manager
Everett T. Ecklund, Design Engineer
Mark Feigenson, Laboratory Assistant
(temporary)
Esther Gordon, Research Assistant
Bennie Harris, Caretaker
E. Kathleen Hill, Secretary 2 ' 1
Janet C. Hunt, Secretary 27
Judith H. James, Typist 28
Willis Kilgore, Jr., Caretaker
Charles A. Little, Electronics Research
Specialist
Niels M. Pedersen, Fiscal Officer
Glenn R. Poe, Electronics Research
Specialist
Elliot M. Quade, Assistant Maintenance
Foreman
Carl M. Rinehart, Instrument Maker
Michael Seemann, Design Engineer —
Mechanical
E. Lynn Smith, Typist, Fiscal Assistant 2 "
James E. Spicer, Maintenance Foreman
Milton T. Taylor, Instrument Maker
-'From January 12, 1976.
-"'From January 26, 1976.
-"Through February 15, 1976.
"From May 6, 1976.
"Through January 15, 1976.
29 Through September 30, 1975.
Hale Observatories
Operated by Carnegie Institution of Washington
and California Institute of Technology
Pasadena, California
Horace W. Babcock
Director
J. Beverley Oke
Associate Director
OBSERVATORY COMMITTEE
Horace W. Babcock, Chairman
J. Beverley Oke, Vice-Chairman
Jerome Kristian
Guido Munch
Allan Sandage
Wallace L. W. Sargent
Maarten Schmidt
Arthur H. Vaughan, Jr.
Carnegie Institution of Washington Year Book 75, 1975-1976
Contents
Introduction 277
Observing Conditions 278
Physics of the Sun 279
Solar magnetic fields 279
Large-scale velocity fields of the sun . . . 280
Big Bear Solar Observatory 280
Coronal holes 280
Surge of 1971 281
X rays from solar flares 281
The lifetime and evolution of fibrils .... 281
Macrospicules and microflares 282
Differential rotation 282
Solar System Studies 283
Jupiter 283
Io 283
Satellites and asteroids 284
Stellar Spectroscopy 284
White dwarfs 284
Studies of proper-motion stars 287
Rotation of B-type stars 288
P- 3 He stars 288
Braking of magnetic Ap stars 288
3 He stars 288
Circumstellar envelopes 289
Stellar spectroscopy at 1 /i,: R Aquarii . . 289
Stellar Chromospheres 289
Absolute luminosities 289
H and K fluxes in main-sequence stars . 289
Infrared and Millimeter-Band Photometry 290
Extragalactic objects 290
Galactic center 290
Nova Cygni 1975 290
1-mm photometry of compact
extragalactic objects 291
Globular Clusters 291
Dynamical models 291
Interstellar Matter and Gaseous Nebulae . 292
High-latitude reflection nebulosities
illuminated by stars of the
galactic plane 292
Interstellar matter 294
Herbig-Haro object in the Orion Nebula 294
The H II region W3 295
Infrared properties of OH masers 295
Dark clouds 295
Mapping of thermal emission from
molecular clouds 296
NGC 7538 296
Pulsars 297
Search for the Hulse-Taylor binary
pulsar 297
X-Ray Sources 297
Supernovae 298
The Galaxy 299
Metal abundances in globular clusters . . 299
Galaxies 299
New redshifts for bright galaxies 299
Distances to nearby galaxies derived
from brightest stars 299
Supergiant cD galaxies 300
Markarian galaxies 300
Seyfert galaxies 301
Infrared observations 301
Nuclear regions of M31 302
Observational search for the
"missing mass" 302
Redshifts, magnitudes, and colors
of a sample of galaxies 302
Chemical history of galaxies 303
Large Magellanic Cloud 303
Population synthesis 304
Variable stars in the dwarf
spheroidal galaxies 304
Image processing of galaxy photographs 305
Ejection from the spiral galaxy
NGC 1097 305
Neighborhoods of galaxies 306
M87 and its globular clusters 308
Clusters of Galaxies 308
Redshifts and magnitudes 308
The Hubble diagram 309
Photometry and redshifts of first
brightest cluster members 310
Brightest cluster galaxies 310
The dynamics of galaxies and
clusters of galaxies 311
Equality of redshifts of E and SO galaxies
in the Virgo cluster 311
A new nearby group 311
X-ray clusters 312
Intergalactic hydrogen clouds 312
Radio Sources 312
3CR identifications 312
Perseus A 312
Quasars and Quasi-Stellar Objects 312
Spectroscopic observations 312
Quasar surveys 313
Search for high-redshift QSOs 313
Absorption lines 313
Optical continuum 315
Emission lines 316
Observational Cosmology 316
The Hubble constant and
deceleration parameter 316
Theoretical Studies 317
Gravitational radiation from pressureless
nonaxisymmetric stellar collapse 317
Gravitational radiation from axisymmetric
stellar collapse with pressure 318
Angular momentum of galaxies
and tidal interaction 318
Instrumentation 318
SIT acquisition and guiding
system (SITEX) 318
Remote operation of Cassegrain
ring rotation 319
SIT-Vidicon digital spectrographs 319
Photon-Counting Image-Intensifier
System 319
Charge-coupled devices 320
Photometry of stellar H and K lines .... 320
Infrared detection systems 321
Astroelectronics Laboratory 321
Guest Investigators 322
Las Campanas Observatory 334
2.5-meter du Pont Telescope 334
Cassegrain camera 335
Cassegrain spectrograph 335
Television system 335
Infrared photometer 336
Coude spectrograph 336
Astronomers' Lodge 336
Bibliography 336
Staff and Organization 341
Research Fellows and Students 341
INTRODUCTION
New data on very faint sources far
out in the Universe are beginning to
accumulate at impressive rates. This
situation, long worked for by observers,
results from years of effort on the part
of many individuals and groups to per-
fect digital or linear photoelectric de-
vices capable of precise photometry on
large numbers of picture elements
within an optical image — in other
words, to count incoming photons in
the picture elements and to present the
resulting large amount of information
in a useful way. Such devices are
equally important in the focal plane of
a telescope or at the output end of a
spectrograph that is attached to the
telescope. But the photographic plate,
especially when employing the most
modern emulsions of high information
content and optimum color sensitivity,
continues to be invaluable, particu-
larly for wide-angle recording.
Two groups at the Hale Obser-
vatories, Drs. Allan Sandage and
Jerome Kristian, using the photo-
graphic method, and Drs. James E.
Gunn and J. B. Oke, using an image
tube, are reporting the discovery of im-
pressive numbers of new faint clusters
of galaxies. Such clusters are of impor-
tance because it is known that the
brightest galaxy in a cluster (for rea-
sons not yet understood) has a standard
luminosity that is very nearly the same
for any cluster. Thus the relative dis-
tances to such clusters can be derived,
and if the redshifts of a number of the
galaxies in each cluster can be meas-
ured with the spectrograph, new points
in the Hubble diagram of redshift ver-
sus distance can be plotted.
Not only are dozens of distant clus-
ters of galaxies being found, but the
production of spectra and the meas-
urement of redshifts of galaxies in
them is proceeding at a record rate. For
this work, Westphal, Kristian, and
Sandage are using an SIT (silicon in-
tensifier target) Vidicon on a prism
spectrograph at the 5-meter telescope,
while Gunn and Oke are using the
SIT- Vidicon with a grating spectro-
graph. Both of these SIT systems have
been derived, with various technical
improvements, from the system de-
scribed in 1971 by Dr. Thomas B.
McCord and Westphal (Year Book 71,
p. 705). Most of the new redshifts being
measured are greater than z = 0.2, and
a substantial fraction have z > 0.4.
These individual galaxies are of small
angular extent, and most are much
fainter than the uniform light of the
sky (airglow and zodiacal light) that
overlies them. Spectroscopy involves
subtracting electronically the spec-
trum of the sky from that of the object
plus sky. The principal limitation now
is not measuring the spectra but find-
ing faint candidates and positioning
them on the slit. Observing depends
upon low-level closed-circuit TV sys-
tems for these operations, which for-
merly involved the much more difficult
technique of making blind offsets from
nearby stars.
A significant new finding by San-
dage further illuminates our under-
standing of the structure of the Galaxy.
A photographic survey of the north and
south galactic polar caps, in progress
for several years, has not only resulted
in discovery of many new remote clus-
ters of galaxies, but has revealed, in
our own Galaxy, faintly luminous
nebulous regions, often in irregular
connective patterns that extend over
distances of degrees (see Fig. 4). The
patterns display much local structure.
Sandage's filter photography shows
them to be reflection nebulae consist-
ing of dust. The local surface bright-
ness of the dust clouds that scatter il-
luminating radiation depends upon
their optical thickness. Sandage shows
that the observations are in good
agreement with the hypothesis that
the dust clouds are diffusely illumi-
nated by general light of stars in the
277
278
CARNEGIE INSTITUTION
galactic plane. The surface brightness
of the denser reflection nebulae is pre-
dicted to be about 25.5 magnitudes per
square arc second. The brightness of
the clouds viewed as reflection nebulae
is directly related to their extinction of
the light of distant galaxies observed
through them; this extinction can be as
high as Am ~ 0.3 even in the direction
of the pole of the Galaxy. Some parts of
the north polar cap are free of detecta-
ble nebulosities to a level of 26 mag per
square arc sec. Studies of the distribu-
tion and surface brightness of irregular
reflection nebulosities in high galactic
latitudes should produce detailed
knowledge of the extinction, leading to
more precise photometric corrections
for the magnitudes of remote galaxies.
This year marks the completion of
basic construction of the 2.5-meter
Irenee du Pont Telescope of the Las
Campanas Observatory in Chile. This
project was initiated in December 1970
by a gift to the Carnegie Institution
from Mr. and Mrs. Crawford H.
Greenewalt. Design of the instrument
by the staff of the Hale Observatories
was begun soon thereafter, following a
resolution by the Trustees that the In-
stitution should augment the funds
available to permit construction of a
telescope of the 100-inch class. The
telescope was constructed in the
United States, shipped to Chile, and
erected at the Observatory under the
direction of Mr. Bruce H. Rule with the
collaboration of Las Campanas person-
nel. In May 1976, optics were installed
and preliminary test photographs ob-
tained. Staff members believe that the
care taken in achieving near-perfection
of the Cassegrain optical system of the
telescope should permit the instrument
to match the excellent quality of the
natural observing conditions that pre-
vail at the site.
The lengthy process of installing and
adjusting the electronic control system
and of perfecting all supplementary
systems required for routine operation
of the telescope is under way. Comple-
tion of these activities, together with
the requirements for adjusting and
testing the first series of auxiliary in-
struments for the telescope, is expected
to require several additional months.
Construction of the large coude spec-
trograph is a separate task that is ex-
pected to require three to four years.
It can now be made known that Miss
Henrietta H. Swope was the donor of a
very substantial gift to the Institution
in 1967. This gift permitted develop-
ment of the Las Campanas Observa-
tory site, with its road and water sys-
tem, as well as purchase of the 1-meter
telescope that was installed in 1971. In
accord with the wish of the donor, this
gift remained anonymous for a number
of years. It is evident that Miss Swope's
generosity was of primary importance
in the establishment of the new obser-
vatory.
OBSERVING CONDITIONS
The 2.5-meter Hooker Telescope at
Mount Wilson was used for observa-
tions on 276 complete nights and 41
partial nights for a total of 2117 ob-
serving hours. Rainfall for the year
was 551 mm, and total snowfall was
1066 mm.
The 5-meter Hale Telescope at
Palomar Mountain was used for a total
of 3489.8 hours, of which 3085.1 were
nighttime hours, as shown in Table 1.
The difference represents twilight time
that was used for infrared or planetary
observations not requiring a dark sky.
Total precipitation at Palomar was 596
mm, with 1168 mm of snow. The
maximum temperature was 35.0°C in
July; minima were -7.8°C in January
and March.
Public visitors at the Palomar Ob-
servatory numbered 133,990 for the
year.
HALE OBSERVATORIES
279
TABLE 1. 5-Meter Observations
Zero
Total
Hours of
Complete
Partial
Observation
Hours
Nighttime
Month
Nights
Nights
Nights
Worked
Observing
July
25
5
1
261.5
239.6
August
29
1
1
303.3
255.9
September
29
1
305.9
301.3
October
26
4
1
345.4
294.0
November
28
2
1
354.3
307.6
December
25
3
3
326.5
289.8
January
28
2
1
382.3
329.4
February
16
4
9
238.1
196.9
March
22
4
5
278.8
247.2
April
19
6
5
214.3
199.7
May
25
3
3
261.2
212.2
June
27
299
35
3
33
218.2
211.5
Totals
3489.8
3085.1
PHYSICS OF THE SUN
Synoptic observations of the sun con-
tinue on Mount Wilson under the
supervision of Howard. Between June
1, 1975, and May 31, 1976, the follow-
ing observations were obtained:
Direct photographs 334
Ha spectroheliograms,
9-meter focus 621
K2 spectroheliograms,
9-meter focus 618
Full-disk magnetograms 341
Integrated-light magnetic-field
measurements 317
Sunspot drawings 330
Sunspot magnetic-field
measurements 327
Solar observations were made on
341 days.
The daily magnetic-field plots are
published monthly in the National
Oceanic and Atmospheric Administra-
tion Bulletin Solar Geophysical Data.
The magnetic synoptic charts are pub-
lished in the Quarterly Bulletin on
Solar Activity of the International As-
tronomical Union. Partial support for
these observational programs comes
from the National Aeronautics and
Space Administration and the Office of
Naval Research.
Solar Magnetic Fields
A study by Howard of magnetic-field
measurements made on the same day
with two different apertures shows a
daily zero-level error in the setting of
the Mount Wilson magnetograph of
about 0.35 gauss (standard deviation).
From the same observations, the scale
(calibration) errors were found to be
about 10%. On the size scale of the
apertures used in this analysis
( — 10,000 km), the polarity mixing of
magnetic elements varied from one
portion of the period covered by these
observations to the other. During the
period of lower flux values, the
polarities are more mixed. It is this
mixing of polarities which, at least to
some extent, is responsible for the
lower flux measurements. Thus the dif-
ference in the measured total flux be-
tween activity maximum and mini-
mum, which is only a factor of 2, is at
least partially caused by a mixing of
positive and negative elements within
the aperture.
Howard, in collaboration with Z.
Svestka, A. S. Krieger, and R. C. Chase
of American Science and Engineering,
Inc., has studied transequatorial co-
280
CARNEGIE INSTITUTION
ronal loops seen in soft x rays from the
A.S.&E. experiment on the Skylab
mission. The loop system was probably
born through reconnection within a
few days after the birth of the second
region. The loops could be seen in soft x
rays for 2V2 days. Transient "sharpen-
ing" of the loops and a striking
brightening of the whole loop system
for about six hours seem to have been
caused by changes in the magnetic-
field configuration in one of the active
regions. The electron temperature in
the loops, equal to 2.1 x 10 (; K in the
quiet phase, increased to 3.1 x 10" K
during the brightening. At the same
time, the loops became twisted while
rising in the corona.
A study of the magnetic sector struc-
ture of the sun has been carried out by
L. Svalgaard, J. M. Wilcox, and P. H.
Scherrer at Stanford and by Howard.
Using interplanetary magnetic-field
data, 454 sector boundaries were iden-
tified, and the corresponding solar
magnetic field configurations were
mapped. The sector boundaries are
clearly visible in the photospheric
magnetic field as north-south polarity
demarcation lines, extending to about
35° latitude on both sides of the
equator.
Large-Scale Velocity Fields of the Sun
The study of the accumulated Mount
Wilson solar velocity data continues by
Howard and H. Yoshimura of the Uni-
versity of Tokyo. Evidence has been
found in several instances for a large-
scale boundary in the rotational veloc-
ity separating high and low velocity
rates. The years preceding the solar ac-
tivity minimum are characterized by
an increase in the equatorial rotation
rate, a striking decrease in the latitude
gradient of the rotation rate at low
latitudes (<40°), and a corresponding
increase in the latitude gradient at
high latitudes. The rotation of the
polar latitudes remains unchanged.
The rotation rate of the equatorial
latitudes is now (May 1976) nearly up
to the sunspot rate (2.9 /jl rad s _1 ). This
is the first time that such rotation
changes have been observed, and it is
the first time that such sensitive
measurements have been available for
this phase of the solar cycle, which is
now very close to minimum.
Big Bear Solar Observatory
The 65-cm vacuum reflector was fi-
nally brought into proper alignment,
and the coude axis was aligned as well.
Excellent photographs were obtained
with a limiting resolution of 0.3", but it
was found that only rarely did the at-
mosphere permit such good results.
Drs. Alan Patterson and Kenneth
Marsh are now resident at Big Bear to
conduct the observational programs.
Coronal Holes
Zirin took part in the Skylab Coronal
Hole Workshop, which met on several
occasions in Boulder, Colorado, to com-
bine the efforts of a number of people
from different subdisciplines toward
understanding the phenomena of co-
ronal holes. Zirin devoted most of his
efforts to trying to detect the presence
of coronal holes in ordinary Ha and
K-line filtergrams. The results of this
study were somewhat indeterminate;
coronal holes clearly coincide with ex-
tensive areas where the chromosphere
is quiet and relatively free of emission.
It was found that if a hole is known to
be present, it is fairly easy to draw in
its boundaries from the Ha and K-line
pictures. However, given an Ha or
K-line picture, it is at present impossi-
ble to determine whether or not there
is a coronal hole in it. One may thus
conclude that the presence of a coronal
hole has little effect on the underlying
chromosphere, and that the chromo-
spheric phenomena we see do not de-
pend on conduction downward from the
corona. (See also the report of William
Adams on solar rotation near coronal
holes.)
HALE OBSERVATORIES
281
Surge of 1971
Zirin completed work on the analysis
of a peculiar solar surge observed in
1971. This surge ejected a great mass of
material that flew across the surface of
the sun and formed a filament several
hundred thousand kilometers away.
The filament lasted for about 20 min-
utes, whereupon the material rose and
returned to its origin. The significance
of this most unusual event is not clear.
It was possible to explain the radio
emission from the surge with a simple
thermal model at 200,000 °K. In addi-
tion, there is a hotter component of
about 10% of the material at a temper-
ature of several million degrees.
X Rays from Solar Flares
Zirin has made a study of nonther-
mal x rays from solar flares, comparing
Big Bear data with data obtained by a
number of different spacecraft. In
every case (about 15 flares), the hard
x-ray flux (greater than 20 keV) meas-
ures the time derivative of Ha inten-
sity. However, after the third hard
x-ray burst ends, the Ha intensity may
continue near the maximum level if a
thermal cloud is formed. This cloud is
then responsible for the soft x-ray
burst. The energy emitted in Ha is
often about 10% the energy of the
nonthermal electrons, which typically
have a hardness index of about 3. The
radio spectrum produced by these elec-
trons was then modeled, using the in-
ferred density and spectrum from the
hard x-ray burst and the observed Ha
area. It is possible to fit the observed
radio-burst data in almost every case
by appropriate use of synchrotron
self-absorption and free-free absorp-
tion, but unfortunately there are many
parameters and the results are not
convincing. Typical values of magnetic
field of several hundred gauss and den-
sity of 10 9 electrons in the nonthermal
plasma are found. It is not possible to
differentiate between the thin foil and
thick foil models of hard x-ray bursts.
It was found that a clear distinction
existed in Ha between thermal and
nonthermal x-ray bursts. The non-
thermal bursts are invariably visible
in offband Ha (about 0.8 A from line
center), while the thermal bursts pro-
duce no emission offband. Thus the line
width of Ha excited by nonthermal
particles is broad compared to that in a
thermal x-ray flare. The profiles in im-
pulsive x-ray bursts are consistent
with the possible momentum transfer
by hard electrons incident on protons.
It is postulated that the initial energy
input in a nonthermal flare occurs
through acceleration of electrons ac-
cording to a power-law spectrum, the
entire input occurring in the first mo-
ments of the flare.
It was found that x-ray bursts with
multiple spikes usually were related to
optical flares with several stages, each
new Ha brightening or ejection corre-
sponding to a new spike in the x-ray
flux.
The Lifetime and Evolution of Fibrils
A detailed study has been made of
the lifetimes and evolution of fibrils in
McMath 12417, using high-resolution
filtergrams in Ha and Ca II K made at
Big Bear Solar Observatory. It was
found that when viewed near disk cen-
ter the lifetime of a fibril is a monoton-
ically increasing function of its
maximum apparent length. This rela-
tionship, together with the form of the
variation of fibril lengths as a function
of time, suggests that fibrils result
from material being impulsively in-
jected into magnetic-field lines at ap-
proximately 30 km s" 1 and returning
under gravity. The lifetimes and ap-
parent lengths of fibrils are then a
function of the inclination of the field
lines only. A study of wavelength scans
through the Ha line confirms that the
apparent extension and retraction of
fibrils represent true mass motion.
282
CARNEGIE INSTITUTION
Macrospicules and Microflares
From careful scrutiny of Big Bear
full-disk Ha time-lapse movies under
high magnification, Tang has found a
class of small flarelike events that
occur in quiet regions at all latitudes,
including the polar regions. These
events are most noticeable at the limb,
where they typically have the appear-
ance of erupting loops or surges. They
are smaller than a supergranule
(heights, 5-15 x 10 :; km; widths, 1-15
x 10 :; km), last for 5 to 10 minutes, and
rise at velocities up to about 100 km
s~'. In the polar regions, the larger of
these eruptions are apparently the so-
called polar-limb surges that have been
observed in Ha at several observatories
since at least as long ago as 1957 (G.
Godoli and F. Mazzucconi, Astrophys.
J., 147, 1131, 1967).
Small surgelike events, termed mac-
rospicules, have been detected and
studied on the polar limbs in EUV
spectroheliograms from Skylab (J. D.
Bohlin et al, Astrophys. J. [Lett.], 197,
L133, 1975; P. L. Withbroe et al., As-
trophys. J., 203, 528, 1976). The Ha
limb events are similar to the EUV
macrospicules in shape, size, motion,
and duration. On this basis, Moore and
Tang (Bull. Amer. Astron. Soc, 7, 423,
1975) proposed that the Ha limb erup-
tions were the Ha counterpart of the
EUV macrospicules and named the Ha
events Ha macrospicules.
Another discovery from Skylab was
the observation of flare events in x-ray
bright points (L. Golub et al., As-
trophys. J. [Lett.], 189, L93, 1974).
X-ray bright points are the x-ray im-
ages of ephemeral active regions: bipo-
lar magnetic regions that are smaller
than supergranules and with lifetimes
usually less than a day. Since x-ray
bright points occur at all latitudes and
are the only observed centers of activ-
ity in the polar regions, Moore and
Tang (Bull. Amer. Astron. Soc., 7, 423,
1975) proposed that macrospicules are
produced by flares in these ephemeral
active regions. Because of their very
small size, these flares are appropri-
ately termed microflares.
The above interpretations proposed
by Moore and Tang were based on indi-
rect evidence rather than on direct
identification of Ha events with either
EUV macrospicules or flares in ephem-
eral active regions. However, more re-
cently Moore, in collaboration with J.
D. Bohlin at the Naval Research Labo-
ratory and L. Golub at American Sci-
ence & Engineering, has found several
direct identifications of Ha macro-
spicules with EUV macrospicules and
with flares in x-ray bright points on the
limb. Thus the identifications proposed
by Moore and Tang have been con-
firmed.
Differential Rotation
Solar differential rotation meas-
urements give a wide variety of results,
depending upon the method of meas-
urement employed, and it appears that
the differences are genuine properties
of the sun, i.e., different rotation trac-
ers actually rotate at different rates.
Long-lived filaments have long been
used as such rotation tracers, but it is
not at all clear that they are represen-
tative of filaments as a class (of which
they compromise only about 5%). A
study of short-lived filaments was
therefore begun in the hope that the
higher quality of the Ha data currently
available would make it possible to
overcome the difficulties inherent in
such an undertaking. The resulting
measurements produced a rotation
curve with a differential term approx-
imately half the size of the correspond-
ing term in the long-lived filament ro-
tation curve.
A variation of the same measure-
ment technique was then applied to a
small sample of filaments bounding the
unipolar magnetic region occupied by
CHI, a large coronal hole observed by
Skylab. This coronal hole had shown
very little differential rotation during
its several-month life span, and the ro-
HALE OBSERVATORIES
283
tation of the underlying unipolar re-
gion was consequently of considerable
interest. Inasmuch as filaments are
known to form preferentially along
field reversal lines, they furnish a pos-
sible tool for measuring the rotation of
the underlying unipolar region, and
this provided the motivation for our
measurements. The resulting rotation
curve was in remarkably good agree-
ment with that of the overlying coronal
hole.
SOLAR SYSTEM STUDIES
Jupiter
The systematic observation of the
variations in time of the intensity of
the SI and SO lines of the (3, 0) band of
H 2 over various regions of the Jupiter
disk {Year Book 73, p. 131) was con-
tinued by Munch in collaboration with
C. Pilcher of the University of Hawaii.
For the purpose, the coude inter-
ferometric scanner of the 2.5-meter
Mount Wilson telescope was used, with
a spectral resolution of 0.074 A and an
angular resolution of 2". Similar meas-
urements were also made on various
regions of the Saturn disk.
Io
With the 150-mm aperture PEPSIOS
spectrometer of the University of Wis-
consin installed at the coude focus of
the 5-meter telescope, Munch, in col-
laboration with Drs. John Trauger and
Fred Roesler of the University of Wis-
consin, has studied the characteristics
of the sodium emission associated with
Io. The field optics (f/40) allowed use of
a maximum entrance aperture 25" in
diameter for a spectral resolving power
of 150,000. A series of 60 profiles of the
D 2 line was obtained when Io was near
both its eastern and western elonga-
tions. The intensity-velocity distribu-
tion of the D 2 emission within 25" of Io
was studied by line profiles obtained
with a nested set of concentric annular
quadrants of various radii. In addition,
sample D 2 scans were obtained from
regions 25" in diameter centered on the
plane of Io's orbit, at distances of 69",
138", and 207" from Io in the direction
away from Jupiter, and at a distance of
69" from Io toward Jupiter. The main
results of the observations can be
summarized as follows:
1. The D 2 emission rate at a distance
of 2.5" from Io has a value, near eastern
elongation, of 80 kilorayleigh (KR),
corresponding to a column density of 2
x 10 +u cm -2 . The emission rate de-
creases exponentially from this value,
with a scale of 3", out to distances from
Io of around 40". Farther out, the rate
of decrease in surface brightness ap-
pears to be slower according to power
law r~" of the distance r from Io, with n
= 1.7.
2. The line profiles for regions cen-
tered on Io are peaked at the value cor-
responding to the Io orbital radial ve-
locity but are asymmetric, with a tail
extending toward the larger velocities
without regard to sign. This high-
velocity tail extends beyond the veloc-
ity of escape from Jupiter. The degree
of asymmetry of the profiles appears to
increase with distance from Io.
3. At distances ^69" from Io in the
direction away from Jupiter, peak
emission occurs at the Keplerian veloc-
ity around Jupiter; but at r = 69" to-
ward Jupiter, peak emission corre-
sponds to the projected Io orbital veloc-
ity. It thus appears that the density of
sodium atoms falls off sharply inside
Io's orbit toward Jupiter.
4. The brightness distribution of D 2
line emission is not symmetric with re-
spect to Io or to the trace of its orbital
plane.
The presence of the K I a 7699 line in
the spectrum of the Io emission region
was discovered by Munch, Roesler, and
Trauger on August 28, 1975, also with
the PEPSIOS spectrometer. Unlike the
Na D 2 line, the profile of the K I line is
284
CARNEGIE INSTITUTION
symmetrical, with a nearly gaussian
form of 3.16 km s" 1 dispersion. The
measured emission rates imply a ratio
of 20 between the column densities of
sodium and potassium. The paths of
line formation for both species are,
however, different, as indicated by
their different line shapes. Since the
lifetime of the neutral K and Na atoms
is determined by collisional ionization
with thermal electrons in the Jupiter
plasmasphere, and the cross section for
this process is about twice as large for
K as for Na, for a plasma temperature
of 100 ev the abundance ratio Na/K
will be less than the ratio of column
densities, and depending on the results
of model calculations under way, it
may be closer to the value correspond-
ing to the earth's crust (4:1) rather
than the solar value (40:1).
The dependence on time of the shape
of the Io emission cloud, expected from
its interaction with the Jupiter mag-
netospheric plasma, has been studied
by Munch in a series of spatial inter-
ferograms. The rings of a 40-mm aper-
ture Fabry-Perot etalon were projected
on the plane of the sky around Io by
means of a collimator and then reim-
aged with an f/1.4 lens on a magnet-
ically focused intensifier. The device
was mounted on the coude spectro-
graph of the 1.5-meter Mount Wilson
telescope in order to obtain a field of
110" diameter. The etalon spacer was
chosen to make the rings of the D 2 and
Di lines coincident at a central order of
interference equal to 7945. Using a
focal-plane occulting disk to block off
the light of Io, it was possible to photo-
graph rings of 35" diameter with an ex-
posure of one hour. The limiting expo-
sure, around two hours, was set by the
amount of scattered continuum light
from Jupiter. The interferograms ob-
tained show clearly the strong concen-
tration of the Na emission toward Io
and toward a plane close to but not
coincident with Io's orbit. The emission
cloud appears to be flattened to a de-
gree increasing with distance from Io,
and its form varies with the magnetic
latitude of Io in the sense that the
emission lines are weaker toward the
Jupiter magnetic equator.
Satellites and Asteroids
Kowal, working under Sargent's
general direction, continued his search
for faint satellites of the outer planets
and discovered a probable fourteenth
satellite of Jupiter. The new object was
photographed with the 1.2-meter
Schmidt telescope on September 30,
October 1, and October 2, 1975. Its vi-
sual magnitude is approximately 21,
and it is estimated to be between 5 and
10 km in diameter. Not enough obser-
vations were obtained to allow the de-
termination of an orbit for this object,
but a heliocentric orbit has been ruled
out. It is hoped that more observations
can be obtained at the next opposition
of Jupiter, in November 1976.
Plates of the outer satellites of Jupi-
ter were also obtained for the U.S.
Naval Observatory in a project to de-
termine better ephemerides of those
satellites.
A search was conducted for new
satellites of Saturn, but none were
found.
On December 27, 1975, Kowal dis-
covered a new Apollo asteroid with the
46-cm Schmidt telescope. This object,
designated 1975 YA, was found on a
film taken for the supernova search
and was moving across the sky at a
rate of 21° per day when first sighted.
The asteroid's orbit has the unusually
high inclination of 64° and a perihelion
distance of 0.9 A.U.
STELLAR SPECTROSCOPY
White Dwarfs
A deeper insight into the physical
processes that separate the degenerate
stars with hydrogen from those with
helium atmospheres has been obtained
by Greenstein, largely with the mul-
tichannel spectrophotometer. With in-
HALE OBSERVATORIES
285
creased numbers of helium stars, it be-
came clear that convection to various
depths must be episodic for degenerate
stars in the range of temperature
25,000° to 12,000°K. No helium-rich
white dwarf atmosphere has been
found hotter than 25,000°, where
helium lines have their maximum
strength. (Only one star, probably near
50,000°, has both ionized helium and
hydrogen in its spectrum.) Among the
cooler helium-atmosphere white
dwarfs, two have been found in which
the lines of ionized calcium appear.
This extraordinary range of excitation
is still compatible with a reduced
abundance of calcium compared to the
opacity-producing elements. Most
cooler white dwarfs with metallic lines
are extremely metal deficient. A work-
ing hypothesis is that degenerate stars
start very hot, with or without hydro-
gen at the surface, but in any case, ac-
crete interstellar matter of normal
composition. In a relatively short time,
the high gravity results in a very com-
plete separation, with hydrogen at the
top, helium below, and the metals still
deeper. Near 25,000° the stability of
this molecular weight gradient is over-
turned by convection, and both helium
and varying small amounts of metal di-
lute the accreted hydrogen for a time.
The helium atmosphere with metals
visible may represent a later stage as
the molecular weight gradient reestab-
lishes itself when convection stops. It is
not obvious that this happens only once
in the life of a white dwarf, nor that it
requires the star to evolve in the direc-
tion mentioned. An alternate view is
that convection may start at a lower
temperature and because of the de-
crease of opacity result in a higher-
temperature object that gradually cools
again.
An important feature of the hydro-
gen white dwarfs has been discovered
in collaboration with Drs. Alec Bok-
senberg, Robert Cars well, and Keith
Shortridge of University College Lon-
don. The Image Photon Counting Sys-
tem was used in the coude spectro-
graph of the Hale reflector to study a
phenomenon discovered earlier by
Greenstein in 40 Eridani B (Fig. 1) and
Wolf 1346. Abnormally sharp cores
exist at the center of the hydrogen lines
Ha and H/3. The efficiency of the IPCS
system permitted observation of ten
hydrogen-atmosphere white dwarfs,
with proof of the existence of sharp
cores in Ha or H/3 for seven of them.
Observations by Greenstein with the
coude image-tube spectrograph at 27 A
mm -1 had already revealed the exis-
tence of sharp Ha cores in five other
white dwarfs of type DA. These cores
arise from the large optical depth at
the center of Ha and the low boundary
temperature of hydrogen atmospheres
caused by Balmer and Lyman line
blanketing. Greenstein and Deane
Peterson of the State University of
New York at Stony Brook predicted
nonlocal thermodynamic-equilibrium
profiles with cores essentially of ther-
mal width (about 15 km s ', mean ve-
locity). If the projected rotational veloc-
ity of a white dwarf is larger than this
thermal velocity (30 or 40 km s" 1 ) the
sharp core would be destroyed. It is
now clear that very slow rotation, cer-
tainly less than 40 km s _I and probably
for some stars less than 10 km s _1 , is
the rule for hydrogen white dwarfs.
The implication is that during the
red-giant phase considerable transfer
of angular momentum between the
center and envelope of the star must
occur. The mass loss of the red giant
should be accompanied by selective loss
of matter of greater than average an-
gular momentum per gram. The brak-
ing of the object, possibly by interac-
tion of the stellar wind with the in-
terstellar medium via magnetic fields,
seems to be very successful. The core is
finally exposed as a hot subdwarf and
eventually a white dwarf of small spe-
cific angular momentum. If the sun
were to shrink homologously to a white
dwarf, the rotational velocities would
be over 500 km s _1 . Since in many
white dwarfs the mass lost is approxi-
mately three quarters of the star's ini-
tial mass, this mechanism is quite effi-
cient.
286
CARNEGIE INSTITUTION
6480 6520 6560 6600 6640 A
CO
15000
12000-
9000
g 6000-
3000
4280 4300 4320 4340 4360 4380 4400 A
Fig. 1. The line profiles of hydrogen in the classical white dwarf 40 Eridani B. These were obtained
with the Hale Telescope at Palomar, using the Image Photon Counting System developed by A.
Boksenberg of University College London. The plots show the photon counts per picture element with
this modern television system. The background slope is caused by the sensitivity changes within the
entire system. Significant is the enormous width of the Hy line, caused by Stark broadening in the
high-pressure atmosphere. The unusual feature of the Ha line, however, is an extremely sharp deep
core, less than 5 picture elements wide in toto. This deep core is produced at the top of the atmosphere
where the pressure is so low that the Stark broadening is minimal. Part of the apparent width is
instrumental in origin, in spite of the original dispersion of 15 A mm -I . The deep core is produced by
effects of nonlocal thermodynamic equilibrium. However, its sharpness proves that the white dwarf
cannot be rotating rapidly, since rotation of as much as 30 km s - ' at the equator would blur the core
out of existence. The estimated rotation is certainly less than 20 km s" 1 and perhaps less than 10 km
s _1 . Thus the period of rotation of the white dwarf must be longer than 30 min and possibly longer
than 1 hr. Were the sun to contract without loss of angular mometum to the size of a white dwarf, its
equatorial velocity would be about 300 km s~'.
HALE OBSERVATORIES
287
As a further result of the sharp line
cores in the hydrogen-line white
dwarfs, it has been possible to derive
rather accurate radial velocities. The
correlation between these and the
prime-focus spectrograph results as
measured by Virginia Trimble of the
University of California at Irvine and
Greenstein is fairly good, with 0.84
correlation coefficient and only small
systematic difference. For the stars
with sharp cores observed either with
the coude image tube or the IPCS, a
mean K-term (i.e., a systematic expan-
sion due to the gravitational redshift)
of +44 km s _1 is derived. This is within
the probable errors of the photographic
value of Trimble and Greenstein, +51
km s" 1 . This confirms a mean mass of
about 0.6 to0.8 9TI o for the white dwarfs.
Individual white dwarfs of special
interest are being searched for by the
multichannel spectrophotometric pro-
gram. In studies made in collaboration
with Boksenberg, the hot star Feige 7
seems almost certain to be a magnetic
white dwarf, showing, without a
Zeeman analyzer, resolved triplet pat-
terns of H/3 and Hy. The 12,000°
helium-rich white dwarf Gill -54 was
found by Greenstein and Boksenberg to
have only the strong, relatively sharp
lines of ionized calcium. Several sus-
pected late-type red degenerates were
also studied for very weak metallic fea-
tures. The difficulty of separating such
metal-poor, helium-rich, cool degener-
ate stars from extremely high-velocity,
metal-poor, hydrogen-rich G or K sub-
dwarfs is extreme. Only parallax and
proper motion will decide.
Greenstein has published data on
189 multichannel spectra observed for
various types of degenerate stars, in
the form of a number of monochromatic
colors and magnitudes. Study of the
correlation of colorimetric properties
and luminosity for stars of known paral-
lax indicates that the degenerate stars
form a relatively narrow sequence with
a spread of about ±0.7 magnitude.
These color-luminosity curves are pre-
sented in the Astronomical Journal
(81, 323-338, 1976) and predict the ex-
pected photometric parallax to be com-
pared with trigonometric parallaxes. A
considerable number of cool degenerate
stars exist with relatively large pre-
dicted parallax but have not yet been
measured. The space motions are those
of the old disk population, although
some Population I and about 9% Popu-
lation II degenerate stars exist. The
number of cool (<7000°) degenerate
stars has been doubled, but this is not
enough for an elementary theory of
cooling. The low number of red degen-
erates probably arises from the rapid
drop in specific heat of the core below
effective temperature 6000°.
There is evidence that Population II
stars more than 1.5 mag subluminous
are rare. When allowance is made for
differential line blanketing, part of this
discrepancy is eliminated. A large gap
remains between the faintest red sub-
luminous star and the faint red degen-
erates.
Studies of Proper-Motion Stars
A systematic program to improve the
relation between luminosity and color
for proper- motion stars has been con-
tinued by Greenstein with the mul-
tichannel spectrophotometer. About
100 stars of known parallax are in-
cluded in this program, and another
300 of large proper motion. Prelimi-
nary indications are that a sublumin-
ous sequence approximately 1 mag
below the main sequence exists in al-
most any color-luminosity plot of paral-
lax stars. With a still insufficient
number of parallaxes for late-type,
very high-velocity stars, this difference
seems to disappear at dM2 or dM3. The
earlier high-velocity M stars are dis-
tinguished by a relatively greater ratio
of hydride to oxide bands. At lower
temperatures, the hydride bands
weaken, and differential line blanket-
ing may disappear. Alternatively,
there may be few or no very high-
velocity stars of late type and low
luminosity. To illuminate the latter
problem, basic for the existence of a
possible galactic halo of low-mass M
288
CARNEGIE INSTITUTION
dwarfs, a systematic program has been
undertaken. The color-luminosity
curves permit calibration of stars of
known multichannel color. Green-
stein's program will measure all Low-
ell stars with proper motion larger
than 1" per year that are brighter than
magnitude 16.4.
Rotation of B -Type Stars
Preston has used 20 A mm -1 spec-
trograms obtained with the Mount
Wilson 2.5-meter telescope to estimate
projected rotational velocities for 180
stars of types B1-B5. This study com-
pletes his survey of the rotational
properties of the bright northern
B-type stars. No new chemically pecul-
iar stars were detected. Higher-
dispersion (9 or 4.5 A mm -1 ) spectro-
grams were subsequently obtained for
some 40 of these stars with u sin i < 40
km s~~' to better define the v sin i dis-
tribution near the origin and to lower
the detection limits for chemical
peculiarities. This effort led to the dis-
covery of only one peculiar star, HR
7467, which exhibits enhanced lines of
P II and P III and no detectable rota-
tional broadening (v sin i< 5 km s ').
As in the case of 3 Centauri A and t
Orionis B, positive wavelength shifts
for the He I lines XX 4921, 5015, and
6678 indicate a gross overabundance of
3 He ( He/ 4 He ~ 2), but unlike 3 Cen A,
HR 7467 does not exhibit lines of Ga II
or Kr II that can be detected at 4 A mm
dispersion.
P -He Stars
The companion of Rigel is a double-
lined spectroscopic binary with a
period of 10 days. Preston has found
lines of P II and P III in the spectrum of
the primary component and a +0.25- A
shift of He I X 6678 that indicates a
He/ 4 He ratio comparable to those
found for 3 Cen A, i Ori B, and HR
7467. The P III/P II and Si Ill/Si II line
ratios for Rigel B (primary) indicate a
spectral type considerably earlier than
B9, as estimated by Sanford, and more
nearly like that of 3 Cen A. Lines of Ga
II and Kr II are not visible on 4 A mm -1
spectrograms. Instead, enhanced lines
of A II are present, including the lead-
ing lines of RMT multiplets 1, 2, 6, 7,
and 10. Thus, of the four P- H stars
known at present, no two are spectro-
scopically alike. Three, curiously, are
members of visual binary systems in
young stellar associations, while one,
HR 7467, is an isolated field star with
MK type (B5 II-III) and (3, c indices
that indicate a late stage of main-
sequence evolution. All four are slow
rotators, but the low incidence of such
stars among slow rotators in the field
sample suggests that slow rotation is
not a sufficient condition for the pro-
duction of the peculiarities of these ob-
jects.
Braking of Magnetic Ap Stars
Hartoog has completed a study of the
rotation of Ap stars in clusters and as-
sociations to look for magnetic braking.
No evidence for magnetic braking was
found. Hot Ap stars seem to rotate
more rapidly than cool Ap stars at all
ages, so the observed decrease in aver-
age rotational velocity with decreasing
temperature must not be an age effect.
It would appear that most of the angu-
lar momentum is lost very early, prob-
ably before the star reaches the main
sequence.
He Stars
Hartoog, using plates from the Hale
Observatories' plate collection, has
identified He in a Sculptoris. The
wavelengths of XX 6678, 4387, and
5015 all indicate about 50% He.
Jugaku and Sargent (Publ. Astron.
Soc. Pacific, 73, 249, 1961) describe
this star as having underabundances of
oxygen and helium and overabun-
dances of strontium, chromium, and
titanium. In this respect a Sculptoris is
quite different from other He stars.
HALE OBSERVATORIES
289
Circumstellar Envelopes
The direct measurement of the angu-
lar diameter of the circumstellar en-
velope around a Orionis (M2 lb) by A.
P. Bernat and D. L. Lambert (As-
trophys. J. [Lett.], 201, L153, 1975) led
Munch and S. Roesler and J. Trauger of
the University of Wisconsin to attempt
a similar measurement for the more
luminous and distant star /x Cephei
(M2 la). The observations were carried
out at the coude focus of the 5-meter
telescope with the 150-mm PEPSIOS
spectrometer. Using both an annular
diaphragm with a central occulting
disk of 2.7" diameter and the com-
plementary aperture with the same
diameter, scans of the K I \ 7699 and
D 2 lines were obtained over a range of
75 km s _1 under excellent seeing condi-
tions. By subtracting one scan from the
other after normalizing them to the
same continuum level, an excess of
radiation was found in the scan corre-
sponding to the annular diaphragm;
this has been ascribed to emission of
the circumstellar shell. The circumstel-
lar D 2 -line emission has an apparent
width at half maximum of 15 km s" 1
and its peak intensity is 10% of the
stellar continuum. Since /x Cep is at a
distance of no less than 1 kpc, it follows
that the radius of its circumstellar en-
velope is of the order of 2000 A.U.
Stellar Spectroscopy at 1 /a: R Aquarii
In the course of reducing spectra
taken in the 1-micron spectroscopy
program, Zirin found that the variable
star R Aqr, which had been identified
as a source of radio emission by P. C.
Gregory and E. R. Seaquist {Nature,
247, 532, 1974), also displayed the Fe
XIII coronal line at 10747 A. The co-
ronal line was positively identified on
two plates obtained in 1971 and 1972.
A new plate obtained in 1976 did not
show this line. On the basis of the co-
ronal line identification, its intensity,
and the absence of the companion
10798 A line, it was possible to make a
rough model of the coronal cloud that
surrounds R Aqr. At a temperature of
1,000,000°K, the product of the cloud
radius and electron density is 2 x 10' 8 ,
and the thermal radio flux at 3.5 cm is
1.44 x 10" 15 R cm mJy. Thus, if the
radius of the corona is less than 4 x
10' 5 cm, the observed flux of 8 mJy can
be obtained with a temperature of 2 x
10" K.
STELLAR CHROMOSPHERES
Absolute Luminosities
Wilson has completed and published
his work on the absolute magnitudes of
over 700 late-type giants, subgiants,
and supergiants from the widths of the
chromospheric K-emission compo-
nents.
H and K Fluxes in Main-Sequence
Stars
The investigation of variations in
flux of the chromospheric H and K
emission in main-sequence stars has
continued. Several objects have com-
pleted about one cycle of variation, and
several others appear close to comple-
tion. These cyclic variations are pre-
sumably the stellar analogs of the solar
cycle. It is anticipated that this work
will be published sometime in the com-
ing calendar year.
290 CARNEGIE INSTITUTION
INFRARED AND MILLIMETER-BAND PHOTOMETRY
Extragalactic Objects
This year marked the conclusion of a
long-term infrared study of several ex-
tragalactic objects carried out by
Neugebauer and Becklin to set limits
on the source of infrared energy by
studying the objects' time history. The
1967 and 1976 light curves for 3C273
and 3C345 are shown in Figs. 2 and 3.
For 3C273 no significant variability
greater than the uncertainties of about
10% is seen at either 2.2 or 3.5 /jl. Since
the infrared energy distribution of
3C273 appears to peak at 3.5 /jl, it can
be concluded that there is no evidence
for infrared variability from this QSO.
For 3C345, 2.2-yu. variations of a factor
of 3 are shown over the years; the vari-
ations appear to have the same phase
and amplitude as visible variations.
Changes by a factor of 2 within one
month's time rule out a simple thermal
source for the 2.2-^t radiation from
3C345.
2.5-/U spectra of several of the 2.2-^u
point sources shown on the high-
resolution maps of Becklin and
Neugebauer (Astrophys. J. [Lett.], 200,
L71, 1975) have been made. Three of
the sources, including the brightest,
show strong carbon-oxide absorption at
2.3 /jl, indicating that they are probably
individual late-type giant or super-
giant stars at the galactic center. Two
sources do not show CO absorption; one
is associated with the brightest 10-yu
emission, while the other is near the
nonthermal radio source found by B.
Balick and R. L. Brown (Astrophys. J.,
194, 265, 1974).
Observations at 10 \x have been
made by Willner. The major result is
that the 10-yu, silicate feature is shown
to be mostly interstellar. Also, the Ne
II line at 12.8 /jl has been mapped at 5"
resolution; the results indicate that the
10-/X emission is not correlated closely
with the ionized gas.
Galactic Center
During the past year much work has
been done on the central 3-pc core of
the center of the Galaxy by Becklin and
Neugebauer. In particular, the 2.1 to
Nova Cygni 1975
During the past year extensive in-
frared observations of Nova Cygni
LJ
o
9.0
CD
<
10.0
3.5^
2.2^
t H
i r
3C273
♦ t +
f + + V ^ i m V ^
H MH Ht
J L
67 68 69 70 71 72 73 74 75 76
TIME
Fig. 2. Photometry of 3C 273 at 2.2 /x and 3.5 tt over the time period 1967-1976.
HALE OBSERVATORIES
291
i r
3C345
visual
2.2/x
V
UJ
Q
Z>
<
♦♦fc.
* it
W
t
\
I »
»♦
67
68
70
71
TIME
72
73
74
75
Fig. 3. Comparison of visual light and 2.2-/U. variability of 3C 345 over the time period 1967-1975.
1975 have been made by Neugebauer,
Becklin, and many members of the in-
frared group. The photometric results
are in basic agreement with those of J.
S. Gallagher and E. P. Ney (As trophy s.
J., 204, L35, 1976) but are much more
extensive, particularly in the early and
later phases of the nova. In addition,
spectroscopy with 1% resolution was
obtained in the 2.2-fx region. The ob-
servations are easily interpreted in
terms of a uniformly expanding shell of
ionized gas. The gas in initially opti-
cally thick at infrared wavelengths and
shows a Rayleigh-Jeans spectrum and
the Brackett y line at 2.16 fx in absorp-
tion. After four days the gas becomes
thin, the infrared spectrum becomes
flat, and the By line goes into emission.
The flux decreased as t~ 2 until —200
days, indicating that the gas is in a
shell rather than uniformly filling an
expanding volume. This dependence is
now becoming closer to a t~' A function,
which indicates that the shell has be-
come more diffuse.
1-mm Photometry of Compact
Extragalactic Objects
This program, headed by M. Werner,
has resulted in the detection of five
sources: 3C84, 3C120, 3C273, 3C279,
and BL Lacertae. The derived fluxes lie
in the range 5-20 janskys. Addition-
ally, 3or upper limits of <5 Jy have
been set for a number of objects, includ-
ing NGC 1068, NGC 4151, and M82.
None of the sources detected show any
significant 1-mm flux over what is ex-
pected from an extrapolation of the
microwave spectrum. The latter limits
are, however, just at the point where
they are becoming significant in dif-
ferentiating thermal from nonthermal
origins for the large amount of infrared
radiation present in these objects. Ad-
ditionally, 3C273 has been monitored
on a bimonthly basis from December
1973 to March 1976 and has shown no
firm evidence for variation greater
than —30% of the mean flux, which is
-13 Jy.
GLOBULAR CLUSTERS
Dynamical Models
Theoretical dynamical models of
globular clusters incorporating veloc-
ity anisotropy and rotation have been
constructed by Gunn to aid in interpre-
tation of his and Dr. R. F. Griffin's
(Cambridge Observatories, England)
292
CARNEGIE INSTITUTION
radial-velocity data. The data for M3
will be analyzed first. The analysis is
not quite complete as of this writing,
but it is clear that models constructed
from central- velocity dispersions alone
are woefully inadequate to describe the
true dynamical state of clusters. M3's
light in its central regions, as
elsewhere, is dominated by giants, and
the radial-velocity data clearly indi-
cate that there is a "cold" (low- velocity
dispersion) population of giants in the
nucleus. Conventional dynamical ar-
guments for systems with relaxation
times as short as those for globular
clusters require that any such popula-
tion have high mass, and it is possible
that these stars arise as the result of
the evolution of close binaries, al-
though no radial-velocity radiation
survives to this epoch.
For M13, rotation is evident, and an
exciting possibility exists of combining
the radial-velocity data with the preci-
sion proper-motion measurements of
Dr. Kyle M. Cudworth at Yerkes Ob-
servatory to find complete space mo-
tions for individual stars. This will not
only give an almost completely unam-
biguous look at the dynamics but will
also yield a distance for the cluster in-
dependent of all astrophysical assump-
tions (except perhaps for 1/r 2 gravity
and the Boltzmann equation).
INTERSTELLAR MATTER AND GASEOUS NEBULAE
High-Latitude Reflection
Nebulosities Illuminated by Stars
of the Galactic Plane
A new survey of the north and south
galactic polar caps has been in progress
for several years by Sandage, using the
1.2-meter Palomar Schmidt with fine-
grain IllaJ and 127-04 high-
information-capacity plates. Its pur-
pose is to find remote clusters of
galaxies with which to extend the
HubBle diagram in a program con-
ducted jointly with Kristian and
Westphal.
During the five years of the survey,
there have been found occasional
high-latitude fields that show bright
nebulous regions, often in connective
patterns that are correlated over dis-
tances of degrees.
A particularly interesting region oc-
curs near the galaxies M81 and M82 at
/ = 142°, b = +41°, shown in the very
high contrast print of Fig. 4. The fila-
mentary nebula of highest surface
brightness at a 5() = 9 h 48 m , S, = +71°
10' is almost due north of M81 and
M82, which are shown as very overex-
posed images near the lower left border
of the print.
The nebulosities over the print form
only part of a very extensive network of
bright structures that extends over the
adjacent Sky Survey field at +72°, 9 h
4 m ; it can be seen on the original Sky
Survey prints.
The fact that regions are nearly
equally visible on the red and the blue
prints, and especially that they are
equally bright on a 103aD emulsion
behind a Wratten-16 filter (XX 5200-
6000 A) where no emission lines occur,
is strong evidence that the structures
are reflection nebulae. The source of il-
lumination at these high-galactic
latitudes is of much interest.
The observed surface brightness of
the regions in Fig. 4 averages SB = 25
mag/D" for the brightest area. Calcula-
tions of the flux due to the light from
the galactic plane show that the appar-
ent magnitude of the plane at any
height is -6.7 V magnitude. The sur-
face brightness of dust that scatters
this light depends on the optical thick-
ness of the structure. But if the same
dust that obscures the background
galaxies by Am also scatters the light
from the galactic plane, then the sur-
face brightness of the nebulosities
should be SB V ~ 24.2 - 2.5 log Am
mag/D".
From galaxy counts on and off the
brightest nebulosity in Fig. 4, and from
HALE OBSERVATORIES
293
i'ijfc
r fi:f,
"&*/.""'» ':■'.'
«'*' t&
Fig. 4. A high-contrast print from part of a Illa-J plate (+ Wratten 4 filter) of a high-latitude region
(a 3 „ = 9 h 49 m , 8 5( | = +71°10', / = 142°, b = +42°) that contains varied reflection nebulae. Overexposed
images of the galaxies M81 and M82 are in the lower third of the print near the left border. The
surface brightness of the brightest reflection nebulae is SB, — 25 mag per square sec, which agrees
with predicted values from a model in which the illumination of the high-latitude dust is provided by
the galactic plane. The scale of the print is 2°45' in width and 3°32' in length.
294
CARNEGIE INSTITUTION
the 21-cm column density of iV H , = 5 x
10 20 atoms cm -2 in the same region, as
measured by Heiles, it is estimated
that Am = 0.3 mag for the dense
nebulae shown. Hence, if the galactic
plane is the source of illumination, the
surface brightness is predicted to be ~
25.5 mag/D". This is in such good
agreement with the observations as to
suggest that the plane is in fact the il-
luminator.
Because the plane is always there,
everywhere, the presence or absence of
faint nebulosities at high latitudes that
have SB less than or equal to —25
mag/D" should act as an indicator of
dust clouds with extinction contrasts of
Am =s 0.3 mag. Hence the presence or
absence of reflection nebulae of this
brightness should aid in new studies of
the extinction through the halo in the
polar directions.
The polar survey is not yet complete
but is far enough along to note that al-
though some fields do have many re-
flection nebulosities at this level and
fainter, many parts of the north polar
cap are free of detectable nebulosities
to SB, ~ 26 mag/D", which means that
the extinction is not lumpy on a scale of
~ 0.2 mag in these regions. But,
equally interesting, the presence of any
reflection nebulae at high latitude
shows the presence of dust above the
galactic plane with optical depths that
can be as high as r = 0.921 Am ~ 0.3.
Interstellar Matter
The study of several cometary reflec-
tion nebulae by Greenstein had shown
peculiarities in the form of differences
in the spectrum between the reflection
nebula and the central object. In col-
laboration with Dr. E. Ye. Khachikyan
of the Byurakan Observatory, a study
of Greenstein's spectra of R
Monocerotis and NGC 2261 showed an
outward acceleration, increasing nega-
tive velocities outward from the central
object. There is also increasing
strength of hydrogen and Ca II. The
nebula is in general somewhat bluer
than the central object and does not
show a steep upturn toward the in-
frared at 1 ix. Greenstein and Oke have
discovered anomalously high reflectiv-
ity near A 6500 in the "Red Rectangle"
centered on the star HD 44179. The
dust grains in the Red Rectangle differ
substantially in physical properties,
therefore, from those in NGC 2261. HD
44179 has an enormous Balmer jump,
requiring a temperature near 8000°,
and surface gravity log g ~ 0. No
available model atmosphere fully ex-
plains the spectrum of HD 44179, al-
though a circumstellar envelope of hy-
drogen can produce a large Balmer dis-
continuity.
Herbig-Haro Object in the Orion
Nebula
Photographs of the Orion Nebula in
radiation of the [O I] A. 6300 line show a
nonstellar double object, 13" in p. a.
280° from the star U 1782 (P. P.
Parenago, Trudy Sternberg Astron.
Inst., 25, 1954), which does not appear
on photographs in recombination lines.
A series of spectra of this object at 29 A
mm _1 dispersion was obtained by
Munch with the f/1.5 camera of the
5-meter telescope coude spectrograph
and an image intensifier. Guiding on
the object was possible through the use
of a two-stage electrostatically inten-
sified eyepiece, filtered to cut down [O
III] and Ha light. The spectra obtained
show the characteristics of classical
Herbig-Haro (HH) objects; namely,
great strength of [N I], [Fe II], [O I],
and [S II] with respect to Ha. The most
remarkable feature of the object, how-
ever, is that its spectral lines are dis-
tinctly double, with one component
shifted by -250 km s _l with respect to
the other, which is intrinsically
brighter in [O I] and which has nearly
the same radial velocity as the mean of
the Orion Nebula. It is conjectured that
the two kinematically separated com-
ponents correspond to the two distinct
nuclei seen in the [O I] direct imagery.
The large shift of the weaker line spec-
trum with respect to the nebula allows
recognition of its [N II] and Ha lines
HALE OBSERVATORIES
295
within the greatly overexposed lines of
the nebula. The intensity ratio be-
tween [O I] A 6300 and Ha in this
high-volocity component is 1.5. The
large relative shift between the two
components of this HH object implies a
time scale of only 50 years for its for-
mation as a double. The violent nature
of the phenomena involved in the evo-
lution and formation of the HH objects
is thus exhibited in a more striking
fashion than heretofore.
The H II Region W3
Further observations of the region
W3, a site of recent active star forma-
tion, have been carried out by S. Will-
ner, a Caltech graduate student who
discovered the presence of an emission
line of Ar III at 8.99 /x. He also meas-
ured more accurately the Ne II and S
IV lines found previously. On the basis
of the 8 to 13-/JL spectrophotometry and
broadband measurements at 5 and 20
fx, the compact sources in NGC 7538
and W3 can be divided into two classes.
The first class, which includes NGC
7538, IRS-1, and W3-IRS-3, consists of
compact sources whose emission from 5
to 20 fx can be described by a single
dust temperature. The second class,
which includes all the other sources in
the two H II regions, consists of both
compact and extended objects that ex-
hibit a range of dust temperatures. The
mass of the emitting dust can be de-
rived for those objects with a single
temperature; for the others, only a
lower limit can be obtained. The mass
of ionized gas can also be deduced for
those objects that have been observed
at radio wavelengths. NGC 7538-IRS-l
has a gas-to-dust ratio of 75; many of
the other objects have upper limits
below 1000. Other workers have found
that the dust inside H II regions is de-
pleted; little evidence for depletion has
been found here.
Infrared Properties of OH Masers
The 1.5-meter telescope at Mount
Wilson, which has recently become op-
erational as a good infrared telescope,
was used by S. Beckwith, a graduate
student at Caltech, in the continued
program related to OH maser sources.
The work was done in collaboration
with N. Evans of Owens Valley Radio
Observatory. Discovery of the infrared
sources associated with OH 1821-12
(E. G. Hardebeck, Astrophys. J., 172,
583, 1972) and OH 21.4-0.7 (Evans e*
at., in preparation), as well as meas-
urements of ON-4 at 10 fj. and 20 /x,
concluded the observations of 1612
MHz, Type II OH maser emitters (W.J.
Wilson, A. H. Barrett, and J. M. Mo-
ran, Astrophys. J., 160, 545, 1971). The
major results of the study are:
1. Of the eight OH maser sources in-
cluded in the study, every one has an
associated infrared source.
2. Of five sources followed for more
than one year, four were found to vary
both in the infrared (2-20 /x) and at 18
cm. In all cases the infrared and radio
fluxes varied in the same sense.
3. One of the infrared sources, that
associated with OH 30.2-0.3, is shown
to have 2A-fx CO band, characteristic
of stars with spectral type later than K.
4. Based on the observed flux at 3.4
[X and 12.5 \x, infrared pumping of the
1612 MHz radio line is possible at 34 fx
but not at 2.8 /x.
Work continued on the properties of
1665 MHz OH masers with the dis-
covery of two more infrared sources as-
sociated with the 1665 MHz masers OH
35.6 and OH 40.6 (Evans etal., in prep-
aration). The infrared sources are very
weak and there appears to be no corre-
lation between near-infrared flux and
1665 MHz line strength in all of the
sources studied. No property character-
istic of infrared sources associated with
these masers has been found.
Dark Clouds
Scanning of dark clouds has con-
tinued by J. Elias, a Caltech graduate
student. To date, three small cloud
complexes have been scanned, those
near IC 5146, NGC 7023, and Sharp-
less 239. In addition, scanning of the
296
CARNEGIE INSTITUTION
two closest large complexes, in Taurus
and Ophiuchus, has continued. Scan-
ning of these regions is from half to two
thirds completed.
The results to date demonstrate that
the sources detected so far — about 10 :i
in number — are for the most part
background objects similar to those de-
tected by earlier, lower-sensitivity sur-
veys. Several interesting objects have
been found. Some photometry and
low-resolution spectrophotometry have
been done on these, but the results are
not yet fully analyzed, and definite
conclusions cannot be drawn.
Mapping of Thermal Emission from
Molecular Clouds
Maps with 1' resolution covering ~
5' x 5' have been made by Werner and
his collaborators of the 1-mm emission
from four sources, W3, Sgr B2, W49,
and M42 (OMC-1). In each case, the
distribution of 1-mm surface bright-
ness shows a sharp peak superposed on
an extended background having < 25%
of the peak brightness. The peak 1-mm
flux density into the 1' beam is 35 Jy
for W3, 90 for Jy for W49, 215 Jy for
OMC-1, and 310 Jy for Sgr B2. These
fluxes are much too high to be due to
free-free emission or unresolved
molecular lines, and the 1-mm radia-
tion is almost certainly due to thermal
emission from dust. Thus it is notewor-
thy that the peaks are seen at the posi-
tions of compact H II regions and/or
pointlike infrared sources, which are
thought from other considerations to be
the youngest and/or most dust-
embedded objects in each of the regions
mapped.
The data have been analyzed in
terms of a simple model of a central
object heating a surrounding dust
cloud in which the dust density de-
creases radially outward as p(r) oc r~".
The best fits to the data are found for
values of n between 1.5 and 2. This
type of density gradient arises very
generally in a variety of gravitational
collapse calculations, so the steep den-
sity gradients inferred from the 1-mm
data may have been produced by the
collapse process that led to the forma-
tion of the central luminosity source.
NGC 7538
The infrared group has continued its
studies of the sites of recent star forma-
tion. One area studied is NGC 7538
(Sharpless 158), an object visible as a
nebulosity on the Palomar Sky Survey.
Aperture synthesis measurements at
21 cm (F. P. Israel, H. J. Habing, and T.
de Jong, Astron. Astrophys., 27, 143,
1973) show a large (5' diameter) H II
region associated with the nebulosity
and a compact H II complex about 3' to
the southeast. A 20-/x map of this com-
pact region shows three infrared
sources within a region about 30"
square, which higher-resolution radio
observations (A.H.M. Martin, Mon. No-
tic. R. Astron. Soc, 163, 141, 1973)
show to be young H II regions. The
most compact of these sources is IRS-1,
which spectroscopy has shown to have
a deep silicate feature.
During the past year, observations
on NGC 7538, centered on the "in-
frared cluster," have been made simul-
taneously at 30, 50, and 100 jx with 1'
resolution from the NASA C-141 Air-
borne Telescope. An exciting aspect of
these observations is the discovery of a
new source not corresponding to any
known radio or near-infrared feature,
located about 1 arcminute to the south-
east of the "infrared cluster," desig-
nated here NGC 7538 (East). The far-
infrared color of this source is similar
to that of the "infrared cluster," but its
luminosity is about 5 times smaller.
Subsequently, a search at near in-
frared wavelengths with the Mount
Wilson 1.6-meter telescope has located
this source, and broadband photometry
shows its 1.6-13 (jl spectrum to be iden-
tical in shape and intensity to that of
IRS-1. However, no radio emission is
seen from this region down to 10 mJy
at centimeter wavelengths (C.G.
Wynn-Williams, private communica-
tion). Size measurements with the
5-meter telescope show that NGC 7538
HALE OBSERVATORIES
297
(East) and IRS-1 are both less than 1"
in size at 10 /x.
Recently a 1-mm map of this region
was constructed by Werner from obser-
vations made during twilight hours
from the prime focus of the Hale
5-meter telescope. At 1 mm the
maximum emission comes from the po-
sition of the "infrared cluster," but a
significant extension is seen that in-
cludes both NGC 7538 (East) and a re-
gion about 1' to the south, which con-
tains an OH maser designated OH(S).
The region around OH(S) shows signif-
icant 100-^t flux but is not seen at 30 or
50 /x, implying that it is extremely cold.
In summary, infrared observations of
the region NGC 7538 between
wavelengths of 1.6 fi and 1 mm have
revealed a wealth of structures and
varying conditions. Detailed analysis
of these data should provide great in-
sight into the evolution of young re-
gions.
PULSARS
Search for the Hulse-Taylor
Binary Pulsar
Kristian, Clardy, and Westphal have
given the first results of a search for
the optical counterpart of the binary
pulsar PSR 1913 + 16, discovered by R.
A. Hulse and J. H. Taylor. The 59 msec
period of this pulsar is the shortest
known except for the Crab pulsar. The
frequency shows cyclic variations with
an 8-hour period, due to an orbital mo-
tion about an undetected second body:
The source is the radio analog of a
single-line spectroscopic binary. It is
the only pulsar known to occur in a bi-
nary system and is of considerable
interest. Among other things, it offers
the possibility of the direct determina-
tion of a pulsar mass and of a "labora-
tory" for a number of new tests of
gravitational theories.
Early synchronous photoelectric
measurements give a limit of V > 23
mag for visible pulsed radiation. This
limit scaled against the Crab pulsar is
roughly the fourth power of the period
ratio. A recent improvement of the
radio position accuracy to 0.2" offers
the possibility of (1) a search for fainter
visible pulsations and (2) detection of
nonpulsed visible light, including that
of the companion, at the pulsar posi-
tion. The Palomar Sky Survey plates
are empty at the radio position, which
sets limits of B > 21 mag and R > 20
mag for any object, pulsating or not,
associated with the pulsar. Rough es-
timates of the distance and absorption
give a limit ofM S3 for the absolute
magnitude of the pulsar and/or the
companion. Deeper data taken with the
SIT area photometer show two faint
red objects 3" to 4" from the pulsar posi-
tion. Further observations of these
stars and a deeper search for pulsed
radiation are anticipated during the
1976 observing season.
X-RAY SOURCES
Oke and Greenstein have made a
study of the x-ray nova A0620-00,
which appeared first as a strong x-ray
source in August 1975. Observations
between September 2, 1975, and March
20, 1976, were made with the mul-
tichannel spectrometer, the SIT digital
spectrograph, or the Boksenberg cam-
era in the coude spectrograph. Infrared
observations at 1.65, 2.2, and 3.5 fx
were also obtained by Neugebauer and
Becklin. The flux below 600 A can be
represented accurately by an energy
distribution somewhere between that
of a model atmosphere and a blackbody
at 25,000°-30,000°K. This radiation
appears to come from an accretion disk
around the x-ray source. One possible
source for the red and infrared radia-
tion is a hot gas at 10"°K, also as-
298
CARNEGIE INSTITUTION
sociated with the accretion disk; in this
case, the density of the gas must be
such that it becomes optically thick in
the near infrared. A second possible
source of the red and infrared radiation
is a cool (4000°K) blackbody, which
must be substantially larger than the
binary system itself. The emission
lines appear to come from the accretion
disk. The derived reddening is consis-
tent with E R _ V = 0.39 obtained from
the Astronomical Netherlands Satel-
lite UV observations. The distance is
approximately 1200 pc. Using the ob-
served data for September 2, 1975, it is
found that the x-ray flux can be pro-
duced by a flow of 1.2 x 10 18 gm s"' of
matter into a neutron star. Using a
simple accretion disk model, this same
flow can generate the energy flux actu-
ally observed from the disk. The model
predicts too small ah energy release in
the outer parts of the disk; this outer
portion of the disk may be heated by
the x rays themselves. The total mass
in the disk is approximately 1.6 x 10 2!
gm, which is much more than the mass
of optically thin gas in the disk and is
of the right order to explain the charac-
teristic times associated with variation
in the optical and x-ray flux.
Oke has completed a study of the
x-ray binary HZ Herculis (Her X-l),
using spectrophotometric observations
around the 1.7-day eclipse cycle.
Analysis of the data at minimum light
yields MIM Q =2.60, L/L Q = 80.7, and
RIR Q = 4.2 for the large primary star,
assuming the standard mass-
luminosity law and conventional stel-
lar evolution. Different but consistent
results can be obtained, provided log g
= 3.60. The hot x-ray heated cap of the
primary star has an angle of 26° and an
effective temperature of 25,000°K.
Analysis of the energy distributions at
phases other than minimum suggests
an accretion disk with a diameter of 2.1
x 10 11 cm; its radiation is consistent
with half of the visual flux coming from
a plasma at 10 ,;o K and N e = 10 14 cnr :;
and the other half from an opaque body
with a temperature of about 20,000°K.
Another x-ray source, A0535 + 26,
which has an x-ray period of 104 s but
no detected optical period, is being
studied by Wade and Oke. Both mul-
tichannel observations and single-
trailed SIT digital spectra were ob-
tained around the 104 s period. The
spectral energy distribution is nearly
flat from 3300 to 10,000 A. Balmer
emission lines are seen and occasion-
ally A. 5876 of He I. No periodicity has
been found so far either in the con-
tinuum or in the emission lines.
SUPERNOVAE
Nine supernovae were discovered in
the course of the Palomar supernova
search during the report year, all with
the 1.2-meter Schmidt telescope. Sar-
gent decided to terminate this part of
the search, which was mainly for
statistical purposes, at the end of 1975.
In all, 178 supernovae have been dis-
covered with the 1.2-meter Schmidt
since the systematic search was begun
by Zwicky in 1959. The search on the
46-cm Schmidt is to go on in an ex-
panded form. The main aims will be to
study the statistics of supernovae in Sc
I galaxies and to maximize the chance
of discovering a really bright super-
nova of Type II.
Kowal completed the work of setting
up photoelectric UBV sequences in 20
of the 1.2-meter supernova fields. Work
is now proceeding on measuring light
curves for several supernovae observed
during the 17 years of the search.
HALE OBSERVATORIES
299
THE GALAXY
Metal Abundances in Globular
Clusters
Searle is making a comparative
study of the chemical history of the
globular cluster systems of the Galaxy,
of M31, and of the Magellanic Clouds.
In this connection, Zinn and Searle are
mapping the spatial distribution of
clusters of different metal abundances
within the Galaxy. A new method of
determining the metal abundances in
individual red giants belonging to a
cluster has been developed. It is based
upon measurements of the integrated
metal-line absorption from spec-
trophotometric scans. This method is
much more sensitive than any previ-
ously used. The observations deter-
mine an intrinsic (reddening-free)
index that varies by a full 1.5 mag over
the abundance range between M71 and
M92.
In this report year Searle and Zinn
have calibrated the new system and
have begun a program to determine the
metal abundances of all globular clus-
ters for which color-magnitude arrays
are available. Particular emphasis has
been placed on remote clusters and
other stellar systems on the fringes of
the Galaxy. So far in this program,
abundances have been determined for
200 stars belonging to 25 systems.
GALAXIES
New Redshifts for Bright Galaxies
A long-range program to obtain red-
shifts for all (—1200) galaxies in the
Shapley-Ames catalogue (whose limit
is m m = 13) in both hemispheres is
nearing completion by Sandage. Dur-
ing the report year, he prepared for
publication new redshifts for 700
galaxies from 1100 plates taken be-
tween 1969 and 1975 with image-tube
spectrograms on the Mount Stromlo
1.9-meter ami the Hale 5-meter tele-
scopes.
All but about 50 Shapley-Ames
galaxies now have measured redshifts,
either from literature sources or from
the new data; so the redshift program
begun in 1935 by Humason and Mayall
is essentially complete.
Sandage and Dr. G. A. Tammann of
the Astronomisches Institute der Uni-
versitat Basel are preparing a redshift
catalog for the complete Shapley-Ames
list, where all available data will be
combined. New two-dimensional mor-
phological classifications of the same
galaxies, using large-scale reflector
plates from Mount Wilson, Palomar,
and Las Campanas, have also been
completed, and these, combined with
the redshift data, are expected to pro-
vide improved luminosity functions for
the various Hubble-van den Bergh
classes.
Distances to Nearby Galaxies Derived
from Brightest Stars
Accurate photometric distances to
nearby resolved galaxies just beyond
the Local Group are still largely un-
known. The brightest resolved stars
provide the easiest fairly accurate
method to find them, short of the very
time-consuming search and photom-
etry of Cepheid variables.
Sandage has begun a long-range
program to use brightest stars to find
distances to many of these resolved ir-
regular dwarf systems whose distance
moduli are less than m — M ~ 28 {D <
4 Mpc).
The first phase of the program is a
strengthening of the absolute mag-
nitude calibration of the brightest
stars. The color-magnitude diagram for
300
CARNEGIE INSTITUTION
the Local Group irregular IC 1613 has
been completed by Sandage and
Katem. The galaxy is at the faint end
of the calibration sample, with an inte-
grated absolute magnitude of only M tt
— -14.4. A special blink survey of
5-meter plates for the reddest stars
shows that the brightest M-type super-
giant has M v = -8.0, M B = -6.0, and
the brightest blue supergiant has M, =
-7.4, M B = -7.6, in agreement with
previous studies of this galaxy reported
in Year Book 69 (p. 95).
Other calibrating galaxies, such as
M33 and the Large and Small Magel-
lanic Clouds, are currently being
searched for the brightest stars on
plates taken with the Mount Wilson
2.5-meter and the Las Campanas 1.0-
meter reflectors.
The strengthening of the present
calibration is expected to take several
years, but the main program for the
unknown galaxies has also begun, with
the first results on the Wolf-
Lundmark-Mellotte (WLM) system
completed by Sandage and Katem. The
color-magnitude diagram from 5-m
plates shows the brightest three blue
stars to have (fi(3)> = 17.85, (V(3)) =
18.00. The brightest red supergiants
have (V) = 18.20. The previous calib-
ration by Sandage and Tammann, done
as a part of their steps toward the Hub-
ble constant, gave, for luminosity class
V galaxies, M fl(3) = -7.99 for the
brightest blue supergiant, and M, (red)
= -7.90, with concordant moduli of (m
- M) AB = 25.95 and (m - M),, =
26.10, for a mean of 26.02 (orD = 1.6
Mpc). This puts the WLM system at the
outer edge of the Local Group.
Plates in good seeing were obtained
with the 5-meter telescope for the addi-
tional dwarf galaxies Pegasus A, Sex-
tans A, Sextans B, Leo A, Leo B, GR 8,
and Ho IX, which is the companion to
M81. Photoelectric calibration of field
stars to V = 17 was completed in each
of these fields, using the Mount Wilson
2.5- and 1.5-meter reflectors. The se-
quences were extended to V = 22,
using 5-m photographs that had sec-
ondary images 5 mag fainter, made
with the Racine wedge calibrator in the
converging prime-focus beam.
It is known that these very nearby
dwarfs partake of the expansion (all
have positive redshifts). The aim of the
long-range program is to determine,
from the distances, the deceleration ef-
fects of the Local Group, and hence to
find its mass from deviations of the
dwarf velocities from the global Hubble
rate.
Supergiant cD Galaxies
Gunn and Thuan are using a new di-
gital spectrograph with a cooled inte-
grating SIT-Vidicon camera at the
Cassegrain focus of the 5-meter Hale
telescope to obtain velocity dispersions
and rotation curves of cD galaxies in
clusters. A synthesis of the total light
of these intriguing systems will also be
performed. This dynamical informa-
tion, along with the light distributions
obtained from plates taken by Thuan
with the Schmidt camera, should per-
mit construction of dynamical models
of the supergiant galaxies.
Markarian Galaxies
Becklin, Neugebauer, Oke, and
Searle have obtained slit spectra in the
visible and spectrophotometric and in-
frared broadband observations of 18
Markarian galaxies with emission
lines. Eight of the program galaxies
can be classified as Seyfert galaxies. In
general, thermal, nonthermal, and
stellar radiation components are pre-
sent. Broadly speaking, one group of
Seyfert galaxies is characterized both
by the presence of a high-density re-
gion of gas and by a continuum domi-
nated by nonthermal radiation. The
continua of the remaining program
Seyferts that do not have a high-
density region of gas are dominated by
thermal radiation from dust and stellar
continuum. The 10 galaxies that are
not Seyfert galaxies are examples of
extragalactic H II regions. There is
HALE OBSERVATORIES
301
evidence for thermal emission at 10 fx
from dust in four of these galaxies.
In a continuing study, Sulentic has
completed a morphological study of
Markarian galaxies 1 through 604,
using the Palomar Sky Survey. The re-
sults have been correlated with radio-
continuum measurements previously
obtained with the National Radio As-
tronomy Observatory 300-foot tele-
scope. It is found that useful coarse
structural information can be obtained
for most of the bright (m =s 15.5) Mar-
karian galaxies from the Sky Survey.
The Markarian galaxies could be dif-
ferentiated into spiral and irregular
morphological types covering a wide
range in absolute luminosity.
Previously Sulentic had suggested
the existence of unusually high radio
luminosities and radio detection frac-
tions for a sample of multiple galaxy
systems. In pursuing this idea, photo-
metric and spectroscopic data are being
obtained for a large sample of these
systems. The correlation of these data
with radio emission and structural
morphology is in progress.
Seyfert Galaxies
Dr. Gregory A. Shields of the Uni-
versity of Texas at Austin and Oke
have obtained and analyzed spectral-
energy distributions for the two Zwicky
compact galaxies ZW 2130+09 (II Zw
136) and Zw 0051 + 12 (I Zw 1), as well
as for 3C273. All of these have strong
Fe II lines in emission. The permitted
line intensities are consistent with a
filamentary emission region having N (
~ 10 9 cm"', R ~ 1 pc, M ~ 10- 3TC , and
v ~ 3000 km s" 1 . The observed H/3 in-
tensities are smaller than expected on
the basis of an extrapolation of the vi-
sual continua to ionizing frequencies,
and this suggests that the emission-
line region does not completely sur-
round the continuum source or that the
continuum slope becomes steeper for A.
< 3000 A. The Fe II line intensities can
be explained by continuum fluores-
cence if the observed line widths result
from radial rather than circular mo-
tion. The derived column density is
~10 2i±1 cm - ' 2 , comparable with the
value implied by the x-ray absorption
in NGC 4151. The O I K 8446 emission
may result from continuum fluores-
cence; alternatively, it can be ex-
plained by Lyman-/3 fluorescence if a
portion of the broad-line region has
t(H«) > 100 as a result of time-
dependent ionization.
Infrared Observations
Persson, J. A. Frogel of the Cerro To-
lolo Inter-American Observatory, and
M. J. Aaronson of Harvard College Ob-
servatory continued their studies of the
stellar populations of elliptical
galaxies by means of infrared photom-
etry. The program of measuring the CO
absorption at 2.2 /jl was largely com-
pleted, and the results are being pre-
pared for publication.
New observational programs were
initiated to investigate further the in-
frared properties of galaxies with the
1.5-meter and 2.5-meter telescopes at
Mount Wilson, the 5-meter telescope,
and the 1-meter telescope at Las Cam-
panas. The first program consists of
measuring the depth of the 2.0-fj. H 2
band in spectra of galaxies and stars.
The band-strength is quite sensitive to
effective temperature in both giant and
dwarf stars later than M2 and offers a
means of studying the relative num-
bers of very late-type stars in galaxies.
A preliminary result of this work is
that the H 2 band is strong in galaxies
and, hence, that a substantial fraction
of the galaxy light at 2 /x is contributed
by M stars. The effect of this parameter
upon stellar synthesis models is under
study.
A second program consists of deter-
mining accurate V - [2.2 /jl] colors for
galaxies. Some preliminary results
were reported by Frogel et al. (As-
trophys. J. [Lett.], 200, L123, 1975).
The importance of these colors follows
from an earlier finding that giant stars
contribute much of the 2.2-^t light of
302
CARNEGIE INSTITUTION
elliptical galaxies. The V - [2.2 fi]
color should thus depend to some ex-
tent upon the morphology of the com-
posite giant branch through the effect
of metal abundance on the luminosities
and effective temperatures of the indi-
vidual stars. A correlation between the
line-blanketing-sensitive U—BorU —
V colors and V - [2.2 /x] should there-
fore exist. The data for three samples
of galaxies, Virgo (30 galaxies), Coma
(12 galaxies), and the field (40
galaxies), show the effect in an unam-
biguous way: A(£/ - V) is approxi-
mately equal to A(V - 2.2 fx]. The V -
[2.2 fi] color thus provides a means of
studying metallicity, and perhaps also
the relative distance moduli of clusters
of galaxies.
Nuclear Regions of M31
Thuan and Oke have obtained spec-
trophotometry measurements of the
nuclear region of M31 under conditions
of excellent seeing and with M31 near
the zenith. A round 3"6-diameter aper-
ture was used for the nucleus meas-
urements, while an annulus with a
diameter of 3" 8- 9" 9 was employed for
the bulge observations. It is found that
the nuclear region with r =s 1'.'8 is only
slightly redder and shows marginally
stronger CN blends that the bulge from
r = 1'.'9 to 4"9. The data can be inter-
preted in terms of an increase in
heavy-element abundance toward the
center of M31 by a factor substantially
less than 2.
Observational Search for the
"Missing Mass"
Thuan used the 1.2-meter telescope
at Palomar to obtain plates of several
edge-on galaxies. They will be used to
test observationally the Ostriker-
Peebles proposal that massive halos
may exist around spiral galaxies and to
study color-gradients in the halos of
galaxies. Photoelectric data will also be
obtained to supplement the photo-
graphic material.
Redshifts, Magnitudes, and Colors of a
Sample of Galaxies
Thuan, in collaboration with G. R.
Knapp of the Owens Valley Observa-
tory and J. Huchra, has started a pro-
gram to obtain redshifts, UBVR
photometry and morphological data for
a complete sample of about 1100
galaxies selected from the Zwicky
catalogue of galaxies and groups of
galaxies, such that m vg =£ 14 m , 6 11 5s
40°, and 8 5= 0°. Since most of the
galaxies (—90%) are spirals, the possi-
bility exists of measuring their red-
shifts via their H I radio emission.
Measurements of a large number of
redshifts are being made quite accu-
rately (—10 km s _1 ) with the sensitive
21-cm system on the 300-foot telescope
at the National Radio Astronomy Ob-
servatory. Redshifts of ellipticals, and
the photometry, are being obtained
with the Palomar 1.5-meter reflector
and the Kitt Peak 91-cm reflector.
The investigations that can be car-
ried out with these data are many, and
all have important implications for
cosmology and for the theories of for-
mation and evolution of galaxies:
1. The study of systematic dif-
ferences (if there are any) in properties,
such as colors between cluster and
noncluster galaxies.
2. The color-magnitude relation
(Sandage, Astrophys. J., 176, 21, 1972)
for elliptical and SO galaxies may be
improved. If the tightness of the corre-
lation holds, it may prove to be a very
powerful cosmological tool.
3. The distributions of colors and ab-
solute magnitudes of galaxies from a
statistically complete sample could be
compared with simple models of star
formation in galaxies (Searle, Sargent,
and Bagnuolo, Astrophys. J., 179, 427 ',
1973).
4. By mapping the local velocity
field, we can apply a "cosmic" virial
theorem to deduce a local mean density
and hope to make headway toward the
outstanding problem of an open or
closed universe.
HALE OBSERVATORIES
303
5. In addition to the radial-velocity
information, the H I data will provide
virial masses and total H I content.
Such a large sample should provide an
interesting statistical study of H I con-
tent as a function of morphological
type.
Chemical History of Galaxies
As part of an investigation into the
comparative chemical histories of the
Galaxy, of M31, and of the Magellanic
Clouds, Searle has begun a program of
classification of the integrated spectra
of their globular clusters. In collabora-
tion with A. Boksenberg of University
College London, spectra were obtained
of 12 globular clusters in M31. These
spectra, obtained with the University
College London image-photon-
counting system, cover the range aa
3500-5000 with a spectral resolution of
5 A. Douglas Rabin, a Caltech
graduate student, is collaborating with
Searle and Boksenberg in the interpre-
tation of these spectra. In a related in-
vestigation, Searle has obtained low-
resolution spectrophotometric scans of
70 of the globular clusters in M31. So
far, all of the observations obtained ap-
pear to be consistent with the idea that
the M31 globular clusters differ from
one another in one parameter only —
presumably the metal abundance.
The situation is essentially different
in the Magellanic Clouds. During the
report year, Searle obtained five-color
photometric observations of the inte-
grated light of 30 Magellanic Cloud
globular clusters at Las Campanas.
These show that there are intrinsic dif-
ferences between the globular clusters
of M31 and of the Galaxy on the one
hand and those of the Magellanic
Clouds on the other. There are also in-
trinsic differences between the clusters
of the Small and Large Clouds.
A possible explanation of the obser-
vations is that in the Galaxy and in
M31 metal enrichment occurred sud-
denly, so that all globular clusters, of
whatever composition, have closely
similar ages; whereas, in the Magel-
lanic Clouds enrichment was slower
and the metal-poor globular clusters in
the Clouds have a substantial range of
age.
Large Magellanic Cloud
The importance of the study of stel-
lar populations in the Magellanic
Clouds, and in particular of the old
stellar component of the LMC disk, has
been previously emphasized. Knowl-
edge of the population structure of the
LMC would make that galaxy, our
closest neighbor, an ideal testing
ground for some aspects of the kind of
synthetic models that must be used
with the unresolved, distant galaxies.
Hardy has studied a number of re-
gions, 3.'4 in diameter each, free of
supergiants, and smooth in the distri-
bution of stars, in and near the bar of
the LMC. Observations were made
photographically in B and V and
photoelectrically in the integrated
light of the UBVR and DDO systems.
His main conclusions are:
The surface-brightness distribution
shows the existence of an exponential
disk of which the bar is a narrow
axisymmetric perturbation. The blue-
length scale for the exponential disk
was found to be a H ~ l = 1.36 ± 0.42 kpc,
while the central surface brightness
was ^ B (0) = 21.93 ± 0.22 mag/D".
There are no significant color gradients
over the sampled baseline of 1.5 kpc,
which roughly corresponds to one
e-folding length over the surface
brightness. But color asymmetries due
to either population or reddening ef-
fects were detected when the bar was
taken as an axis of symmetry.
Color-magnitude diagrams to M v =
0.5 were constructed, and it was found
that the observed giant branch resem-
bles that of intermediate Population II
clusters such as M4 or M75. The sur-
face colors, which include new observa-
tions in the R region obtained at Las
Campanas, are compatible with this
interpretation. There are indications of
304
CARNEGIE INSTITUTION
the existence of a horizontal branch of
Population II, but an important con-
tribution from intermediate-age stars
cannot be ruled out only on the basis of
the colors and the color-magnitude
diagrams.
Luminosity functions were con-
structed by Hardy and then normalized
to the total, light contained in the area
counted (count- brightness-ratios:
CBR). A method was devised that in-
volved comparison of the observed CBR
in both V and B with those of known
globular clusters of different metallic-
ity to determine the fraction of the
total light in Population II stars. It was
found that if the giant branch is inter-
preted as belonging to an M4-M75
type of population, then Population II
contributes 70% ± 10% of the total
light in V and 80% ± 20% in B in the
regions studied, thus leaving only a
small fraction of the light to be ac-
counted for by stars of the solar-
neighborhood type that were too faint
to be detected photographically.
The alternative, that the LMC disk
is a scaled version of the galactic disk,
was discussed. It was shown that the
observed CBR in B light did not fit the
CBR derived from the van Rhijn func-
tion for the solar neighborhood, and it
was argued that this was not only an
effect on the young stellar population
but involved the old component as well.
Alternatives for the formation and
evolution of the LMC disk were consid-
ered. Hardy concluded that the ob-
served disk is likely to be as old as the
galactic disk but, being more metal
poor, resembles an enriched globular-
cluster population rather than an
M67-NGC 188 population, as in the
galactic disk.
New data obtained with the 4-meter
telescope at Cerro Tololo Inter-
American Observatory in January
1976 are now being analyzed.
Population Synthesis
Turnrose has determined the stellar
content of the nuclear regions of seven
nearby Sc galaxies by means of popula-
tion syntheses, utilizing narrow-band
spectrophotometry covering the
wavelength range XX3300- 10,400. A
linear-programming fitting technique
was used in conjunction with simple
models of star formation to provide
physically consistent population mod-
els incorporating multiple generations
of stars. Evidence is found for substan-
tial intrinsic reddening in the nuclear
regions studied. Upper-main-sequence
stars are significant contributors to the
light in most cases, and the lower main
sequence contributes significantly in
all cases. All available indicators are
consistent with evolved- M-star domi-
nance at the longest wavelengths and
low photometric mass-to-light ratios
(0.3 < MIL r < 10). Using absolute
measurements of the emission-line
spectra of the nuclear regions, it is
found that the O-B stars, which arise
naturally in the population models, are
just sufficient to provide the observed
nuclear ionization in all cases except
that of NGC 5194, which may be colli-
sionally ionized. The inferred stellar
populations are largely consistent with
simple evolutionary models of star
formation, using a "local" initial mass
function and a variety of time depend-
ences ranging from exponential decays
with time constants —3-6 billion years
to constant rates. A possibly significant
correlation between nuclear stellar
content and overall dynamical prop-
erties among four of the program
galaxies is pointed out.
Variable Stars in the Dwarf Spheroidal
Galaxies
A rare class of variable stars that
seem to occur preferentially in the
dwarf spheroidal galaxies has been the
subject of two investigations by Zinn in
the current report year.
Together with Searle, he obtained
spectrophotometric scans of five of
these peculiar Cepheids in the Draco
system. Comparing these with similar
observations of RR Lyrae variables in
this system, Zinn and Searle have
HALE OBSERVATORIES
305
found that the masses of the anomalous
Cepheids are from two to three times
the masses of the RR Lyrae stars. This
result, which was unexpected in an old
stellar system whose evolved stars
were believed to have the same mass,
confirms the hypothesis put forward by
J. Norris of Yale University and Mt.
Stromlo-Siding Spring Observatory
and Zinn on the basis of theoretical ar-
guments. How the anomalous Cepheids
acquired their large masses remains
something of a puzzle, however. Either,
as Norris and Zinn argued (see Year
Book 74, p. 338), star formation oc-
curred recently in Draco or, as
suggested as an alternative by Zinn
and Searle, the large masses were pro-
duced by the fusion of the components
of close binary systems.
Until recently, the anomalous
Cepheids had been found in only the
dwarf spheroidal galaxies and the
Small Magellanic Cloud. This past
year Zinn and Conrad C. Dahn of the
Naval Observatory showed that Vari-
able 19 in the globular cluster NGC
5466 obeys the same period-luminosity
law and has the same mass as the un-
usual Cepheids in the Draco system.
The as yet unidentified characteristic
of the dwarf spheroidal galaxies, the
Small Cloud and NGC 5466 that dis-
tinguishes them from other systems is
clearly a key to understanding the ori-
gin of these peculiar variables.
Image Processing of Galaxy
Photographs
There has always been interest in
extracting maximum information from
astronomical photographs. In an ex-
perimental project that utilizes the
capabilities of modern digital com-
puters, Jean Lorre of the Jet Propul-
sion Laboratory's Image Processing
Laboratory and Arp have collaborated
in the analysis of a number of galaxy
photographs. The photographs were
made by Arp on fine-grain plates and
under especially good observing condi-
tions with the 4-meter reflector at
Palomar. The plates have been scanned
with a PDS microphotometer and pro-
cessed by a variety of mathematical al-
gorithms by Lorre at JPL. Techniques
have been demonstrated that (1) en-
hance the contrast and clarity of
nebulous features, (2) detect previously
invisible features, (3) moderately im-
prove resolution (through partial see-
ing compensation), (4) add information
from different plates to improve
signal-to-noise ratios, and (5) compare
response at different wavelengths in
order to portray color differences.
In certain well-known and interest-
ing groups of galaxies, new features
were detected by these procedures. For
example, in the region of NGC 7331
and Stephan's Quintet, luminous
nebulous filaments were discovered
(see Astrophys. J., 210, No. 1, 1976,
November 15). Additional detail is also
revealed in some of the galaxies in
Seyfert's Sextet. In one of the most
striking results, a photograph of the
famous jet in the spherical galaxy M87,
taken with the 5-meter telescope on
fine-grain emulsion, has been pro-
cessed for maximum resolution. The
deconvolution algorithm has reduced
the smallest image sizes by about a fac-
tor of 2 and compressed the dynamic
range so that the jet is seen in this pic-
ture as consisting of a very small well-
spaced and well-aligned series of
luminous points emerging from the nu-
cleus of M87 (see Fig. 5a, b).
Ejection from the Spiral Galaxy
NGC 1097
While cataloging peculiar southern
galaxies in Edinburgh, using plates
from the U.K. 1.2-meter Schmidt in
Australia, Arp learned of the discovery
by R. D. Wolstencroft and W. J. Zealey
of luminous jets emerging from the
bright southern spiral galaxy, NGC
1097. A few months later, with the new
4-meter telescope at Cerro Tololo in
Chile, he obtained a series of deep
plates on this object which reveal that
it is one of the clearest examples of
ejection from a galaxy that has so far
been encountered. The 4-m plates show
306
CARNEGIE INSTITUTION
Fig. 5a. A photograph of the jet in M87 with the 5-meter telescope on fine-grain Illa-J emulsion.
This photograph has been subjected to an image deconvolution algorithm that improved the obtaina-
ble resolution by about a factor of 2 and increased the dynamic range.
that the three jets all point exactly
back to the nucleus of the galaxy and
suggest, therefore, that material has
been ejected from the nucleus.
Neighborhoods of Galaxies
A two-year collaborative project was
initiated by Arp with Francesco Ber
HALE OBSERVATORIES
307
Fig. 5b. The resulting image-processed picture shows the well-known jet resolved into a series of
evenly spaced, well-aligned points of light.
tola and Cesare Barbieri of Padova Ob-
servatory. Its purpose is to study the
contents of fields in the vicinity of
bright galaxies. Ninety-nine bright
spiral galaxies between magnitudes 10
and 12 have been selected, as well as 99
control fields carefully chosen to be
more than 6° distant from any bright
galaxies. The region within 1° of each
of the spirals has been carefully
studied in order to catalog companion
galaxies and disturbed or nonequilib-
rium galaxies, including interacting
double galaxies in which a quantita-
308
CARNEGIE INSTITUTION
tive measure of relative separation
may be recorded. Jack W. Sulentic,
postdoctoral Hale Observatories Fel-
low, has completed the initial survey.
The same kind of census has been com-
pleted for each of the associated control
fields.
In May 1976, Dr. Graziella de Tullio
of the Padova Observatory visited
Pasadena to work on this same prob-
lem, comparing her independent
cataloging of the fields with Sulentic's.
Now a program is under way in both
Pasadena and Padova to obtain photo-
metric magnitudes, colors, spectro-
scopic classifications, and redshifts of
these objects with the Palomar 1.5-
meter and the Asiago 2.3-meter reflec-
tors. A secondary phase of this program
is to determine whether there are any
photometric or spectroscopic dif-
ferences that distinguish companions
of the bright spirals from the back-
ground galaxies in the control fields.
With A. G. Willis and H. de Ruiter of
Leiden Observatory, Arp has com-
pleted optical identifications of accu-
rate Westerbork radio positions around
a number of bright galaxies. Some of
the more interesting identifications in-
clude a new 16.5-17.0 mag object of the
BL Lacertae type very near the large
spiral galaxy NGC 6503, a quasar that
brightened almost 4 magnitudes in less
than one month, and a number of
quasars with redshifts greater than 2,
including one with z = 3.19 (see Year
Book 74, p. 333).
Sulentic and Arp have obtained U, B
double-image plates of a number of
their galaxy and control fields. These
plates permit selection of ultraviolet
excess stars that are quasar candidates
and will allow an independent deter-
mination on the value of radio-quiet
quasar numbers in the general field.
M87 and Its Globular Clusters
Sargent collaborated with C. R.
Lynds of Kitt Peak National Observa-
tory, F. D. A. Hartwick of the Univer-
sity of Victoria, and A. Boksenberg and
R. F. Carswell of University College
London in an attempt to measure the
dispersion in velocity of the system of
globular clusters around M87. The ob-
servations were made with Boksen-
berg's IPCS detector on Lynds' "Gold"
spectrograph at the Cassegrain focus of
the Kitt Peak 4-meter telescope. Be-
cause of indifferent seeing conditons, it
proved possible to obtain a good spec-
trum of only one suspected cluster
(IV-122 in an unpublished catalog by
D. A. Hanes of David Dunlap Obser-
vatory). This object, which has B =
20.6 mag, proved to be a globular clus-
ter (based on its redshift). Remarkably,
IV-122 turns out to be more metal-rich
than the galactic globular cluster M71,
despite the fact that it is in the halo of
M87, about 34 kpc from the center.
This object was measured in a 3-hour
exposure. It thus appears to be feasible
to carry out the original project, given
good seeing. A further attempt on the
problem will be made on the Hale Tele-
scope in 1977.
In view of the poor seeing during the
run at Kitt Peak, the above team de-
voted a considerable effort to M87 it-
self. Lengthy exposures were made to
determine the radial dependence of the
velocity dispersion of the stars in M87
out to 150" from the center. Usable
data were obtained despite the fact
that at this radius M87 is about 1 mag
fainter than the sky. They are now
being analyzed. The ultimate aim of
the observations of M87 and its clus-
ters is to measure the total mass of the
galaxy and its radial distribution.
CLUSTERS OF GALAXIES
Redshifts and Magnitudes
Kristian, Sandage, and Westphal are
continuing a program to extend the
Hubble diagram to redshifts greater
than 0.2. The main candidate lists in
the first stage are (1) the fainter Abell
clusters, (2) clusters found on a special
high-latitude 1.2-meter deep survey
with Illa-J and 127-04 plates, and (3)
HALE OBSERVATORIES
309
faint clusters associated with 3C radio
sources. Redshifts are being measured
with the SIT-prism digital spectro-
graph and BVR magnitudes with an
S20 photomultiplier and the SIT area
photometer. Redshifts have been
measured for 80 clusters, including 60
with z > 0.2, 15 with z > 0.3, and 3
with^ > 0.5 (3C330 at^ = 0.53, 3C123
at z = 0.63, and 3C343.1 at z = 0.75
[preliminary!). With observations by
Spinrad, using the image-dissector
scanner at Lick Observatory, by Gunn
and Oke, using the multichannel scan-
ner, and those reported here, the 20-
year-old "wall" atz = 0.2 has collapsed,
and redshifts to 2Vz or 3 times larger
distance are now routine. The principal
limitation at present is not measuring
the spectra but finding faint candidates
and positioning them on the spectro-
graph slit. It now appears likely that
about 100 clusters in the important
range z = 0.3-0.5 can be measured in
the next few years. Larger redshifts re-
quire rare nights of exceptionally good
seeing and are likely to come more
slowly.
Analysis of some early results in the
range z = 0.2 to 0.3 shows several fea-
tures: (1) In the absence of evolu-
tionary corrections, the Friedmann
models with q u — 1 continue to fit the
data reasonably well. (2) The cosmic
dispersion of 0.25 mag in the absolute
magnitude of the brightest member
and cluster richness found earlier by
Sandage and Hardy is confirmed but is
remarkably small over a large range of
richness. This implies that if the mag-
nitude of the brightest member is gov-
erned by a general luminosity function,
then the slope of the function at the
bright end must be very steep: d log
f(M)ldM 5= 5.
Preliminary analysis of 35 addi-
tional clusters to z > 0.5 supports the
earlier conclusions and begins to show
the expected turnover to bluer B — V
colors at large redshift, as the pass
bands move into the ultraviolet regions
of the galaxy rest spectrum. This color
change is expected to be a useful dis-
covery discriminant, and a program
has been started to use it to find ex-
tremely distant clusters.
The Hubble Diagram
The survey by Gunn and Oke with
the 90-mm image tube and the 5-meter
Hale Telescope now consists of 250-300
plates. On these plates approximately
60 faint clusters have been found. The
south galactic hemisphere sample is
essentially complete, but considerable
work needs to be done on one or two of
the spring fields, where progress has
been slow due to bad weather. Since
their first analysis of the Hubble dia-
gram, an additional 38 galaxies in
faint clusters have been observed, and
redshifts and energy distributions now
exist for most of these. A substantial
fraction of these 38 objects have red-
shifts of 0.40 or more. The SIT spec-
trograph with on-line data handling
and reduction system is now fully op-
erational and is being used to obtain
spectra of very faint galaxies; its ex-
pected speed advantage over the mul-
tichannel spectrometer is realized.
In their first analysis of the Hubble
diagram, Gunn and Oke found that a
significant uncertainty was introduced
because of poor statistics for the rela-
tively close clusters. This has been rec-
tified by work with Thuan on a com-
plete sample of nearby Abell clusters
(all with distance class «s 4, richness
5=1, and |6 | > 30°, 102 in all). Photom-
etry with an intermediate-band sys-
tem, designed to be easily tied to the
spectrophotometric system used for the
distant objects with the 5-meter, has
been carried out on the Palomar 1.5-
meter telescope, and redshifts have
been obtained, mostly with the image-
tube spectrograph and more recently
with the SIT digital spectrograph on
the 5-meter telescope. This sample
should provide very good statistics at
the brighter end of the Hubble dia-
gram. Since the sample is in some
sense complete, it also allows more
careful statistics to be done on sys-
tematic effects with richness and mor-
phological type than have been possi-
310
CARNEGIE INSTITUTION
ble before. Some of this work is being
undertaken in collaboration with J. G.
Hoessel, a graduate student doing
thesis research 
TVNews
OpenLibrary