Full download ebooks at https://ebookmeta.

com

Pattern Recognition and Artificial Intelligence

International Conference ICPRAI 2020 Zhongshan
China October 19 23 2020 Proceedings Yue Lu

For dowload this book click link below

https://ebookmeta.com/product/pattern-recognition-and-
artificial-intelligence-international-conference-
icprai-2020-zhongshan-china-october-19-23-2020-proceedings-
yue-lu/

OR CLICK BUTTON

DOWLOAD NOW
More products digital (pdf, epub, mobi) instant
download maybe you interests ...

Pattern Recognition and Artificial Intelligence Third

International Conference ICPRAI 2022 Part 1 Mounîm El
Yacoubi

https://ebookmeta.com/product/pattern-recognition-and-artificial-
intelligence-third-international-conference-
icprai-2022-part-1-mounim-el-yacoubi/

Artificial Intelligence and Security 6th International

Conference ICAIS 2020 Hohhot China July 17 20 2020
Proceedings Part I Xingming Sun

https://ebookmeta.com/product/artificial-intelligence-and-
security-6th-international-conference-icais-2020-hohhot-china-
july-17-20-2020-proceedings-part-i-xingming-sun/

Advances in Guidance Navigation and Control Proceedings

of 2020 International Conference on Guidance Navigation
and Control ICGNC 2020 Tianjin China October 23 25 2020
1st Edition Liang Yan
https://ebookmeta.com/product/advances-in-guidance-navigation-
and-control-proceedings-of-2020-international-conference-on-
guidance-navigation-and-control-icgnc-2020-tianjin-china-
october-23-25-2020-1st-edition-liang-yan/

HCI International 2020 Late Breaking Papers

Multimodality and Intelligence 22nd HCI International
Conference HCII 2020 Copenhagen Denmark July 19 24 2020
Proceedings Constantine Stephanidis
https://ebookmeta.com/product/hci-international-2020-late-
breaking-papers-multimodality-and-intelligence-22nd-hci-
international-conference-hcii-2020-copenhagen-denmark-
Information management and big data : 7th Annual
International Conference, SIMBig 2020, Lima, Peru,
October 1-3, 2020, Proceedings Coll

https://ebookmeta.com/product/information-management-and-big-
data-7th-annual-international-conference-simbig-2020-lima-peru-
october-1-3-2020-proceedings-coll/

HCI International 2020 Late Breaking Papers Cognition

Learning and Games 22nd HCI International Conference
HCII 2020 Copenhagen Denmark July 19 24 2020
Proceedings Constantine Stephanidis
https://ebookmeta.com/product/hci-international-2020-late-
breaking-papers-cognition-learning-and-games-22nd-hci-
international-conference-hcii-2020-copenhagen-denmark-
july-19-24-2020-proceedings-constantine-stephanidis/

Brain Function Assessment in Learning Second

International Conference BFAL 2020 Heraklion Crete
Greece October 9 11 2020 Proceedings Claude Frasson

https://ebookmeta.com/product/brain-function-assessment-in-
learning-second-international-conference-bfal-2020-heraklion-
crete-greece-october-9-11-2020-proceedings-claude-frasson/

Machine Translation 16th China Conference CCMT 2020

Hohhot China October 10 12 2020 Revised Selected Papers
Junhui Li (Editor)

https://ebookmeta.com/product/machine-translation-16th-china-
conference-ccmt-2020-hohhot-china-october-10-12-2020-revised-
selected-papers-junhui-li-editor/

Emerging Technologies in Computing Third EAI

International Conference iCETiC 2020 London UK August
19 20 2020 Proceedings Mahdi H. Miraz

https://ebookmeta.com/product/emerging-technologies-in-computing-
third-eai-international-conference-icetic-2020-london-uk-
august-19-20-2020-proceedings-mahdi-h-miraz/
 Yue Lu · Nicole Vincent ·
Pong Chi Yuen · Wei-Shi Zheng ·
Farida Cheriet · Ching Y. Suen (Eds.)
LNCS 12068

Pattern Recognition
and Artificial Intelligence
International Conference, ICPRAI 2020
Zhongshan, China, October 19–23, 2020
Proceedings
Lecture Notes in Computer Science 12068

Founding Editors
Gerhard Goos
Karlsruhe Institute of Technology, Karlsruhe, Germany
Juris Hartmanis
Cornell University, Ithaca, NY, USA

Editorial Board Members

Elisa Bertino
Purdue University, West Lafayette, IN, USA
Wen Gao
Peking University, Beijing, China
Bernhard Steffen
TU Dortmund University, Dortmund, Germany
Gerhard Woeginger
RWTH Aachen, Aachen, Germany
Moti Yung
Columbia University, New York, NY, USA
More information about this series at http://www.springer.com/series/7412
Yue Lu Nicole Vincent
• •

Pong Chi Yuen Wei-Shi Zheng

• •

Farida Cheriet Ching Y. Suen (Eds.)

•

Pattern Recognition
and Artificial Intelligence
International Conference, ICPRAI 2020
Zhongshan, China, October 19–23, 2020
Proceedings

123
Editors
Yue Lu Nicole Vincent
East China Normal University Paris Descartes University
Shanghai, China Paris, France
Pong Chi Yuen Wei-Shi Zheng
Hong Kong Baptist University Sun Yat-sen University
Kowloon, Hong Kong Guangzhou, China
Farida Cheriet Ching Y. Suen
Polytechnique Montréal Concordia University
Montreal, QC, Canada Montreal, QC, Canada

ISSN 0302-9743 ISSN 1611-3349 (electronic)

Lecture Notes in Computer Science
ISBN 978-3-030-59829-7 ISBN 978-3-030-59830-3 (eBook)
https://doi.org/10.1007/978-3-030-59830-3
LNCS Sublibrary: SL6 – Image Processing, Computer Vision, Pattern Recognition, and Graphics

© Springer Nature Switzerland AG 2020

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the
material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information
storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now
known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book are
believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors
give a warranty, expressed or implied, with respect to the material contained herein or for any errors or
omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
 Preface

Message from the Honorable Chair and General Chair

Welcome to the proceedings of the Second International Conference on Pattern

Recognition and Artificial Intelligence (ICPRAI 2020). This conference follows the
successful ICPRAI 2018, held in Montreal, Canada, in May 2018. It was organized by
CENPARMI (Centre for Pattern Recognition and Machine Intelligence) at Concordia
University, Montreal, Canada, with the co-operation of the Chinese Association of
Science and Technology, China, and other universities from China and abroad. The
Organizing Committee of ICPRAI 2020 consisted of well-known experts from all six
continents of the world with a dream team of eight world-renowned keynote speakers
to cover the new PRAI frontiers in depth and in breadth with insights. The technical
program included 74 presentations in 5 tracks related to handwriting and text pro-
cessing, computer vision and image processing, machine learning and deep networks,
medical applications, forensic science and medical diagnosis, features and classification
techniques, and various applications. In addition, a workshop on Social Media Pro-
cessing was also organized.
Pattern Recognition and AI techniques and systems have been applied successfully
to solve practical problems in many domains. In co-operation with CENPARMI
members and colleagues at other institutions, ICPRAI 2020 is co-sponsored by the
Chinese Association of Science and Technology and numerous universities and
industrial companies. It brings together a large number of scientists from all over the
world to present their innovative ideas and report on their latest findings, as well as the
fruitful results of numerous CENPARMI members and their students in handwriting
recognition, forensic studies, face recognition, medical imaging, deep learning, and
classification techniques. In addition to the rich technical program, our conference also
featured practical sessions to encourage author-editor and inventor-investor interac-
tions, and social events for all conference participants.
For all the above activities, we would like to thank all those involved for a superb
technical program and conference proceedings: conference co-chairs, Edwin Hancock
and Patrick Wang; track co-chairs, Yue Lu, Nicole Vincent, Pong Chi Yuen, Farida
Cheriet, and Wei-Shi Zheng; workshop co-chairs, Kam-Fai Wong and Binyang Li;
special session chair, Olga Ormandjieva; as well as the publication and publicity chairs
and their committee members.
We would like to express our gratitude to the numerous committee members for
taking care of financial and sponsorship matters (Louisa Lam, Wen Song, Marleah
Blom, Mounîm El Yacoubi, and Jun Tan). Special thanks go to local arrangements
of the Chinese Association of Science and Technology and the government officials of
Zhongshan City, and Nicola Nobile who dedicated himself to a great variety of duties,
ranging from conference secretariat business, publications, software for handling paper
submissions, reviews to selections, plus numerous other chores. They, together with
vi Preface

CENPARMI members and students, created a very warm and comfortable environment
to work in.
Thanks are also due to the organizations listed in the proceedings and those of the
organizers of ICPRAI 2020, and the administration of Zhongshan City.
Finally, I hope you found this conference to be a rewarding and memorable
experience. We hope you enjoyed your stay in the beautiful Zhongshan City, a hub in
the Guangdong, Hong Kong, Macau Greater Bay Area, and the birthplace of great
historical figure Dr. Sun Yat-sen.

August 2020 Ching Y. Suen

Yuan Y. Tang
 Organization

Organizing Committees

ICPRAI 2020 was hosted by Zhongshan City and organized by CENPARMI (Centre
for Pattern Recognition and Machine Intelligence) of Concordia University, Canada,
and journal editors and scientists:

Honorary Chair

Ching Y. Suen, Canada

General Chair

Yuan Y. Tang, Hong Kong, China

Conference Co-chairs

Edwin Hancock, UK
Patrick S. Wang, USA

Special Sessions and Workshops Committee

Adam Krzyzak, Chair, Canada

Camille Kurtz, France
Cheng-Lin Liu, China
Khoa Luu, USA
Takeshi Masuda, Japan
Kam-Fai Wong, Hong Kong, China
Vera Yashina, Russia

Program Chairs

Farida Cheriet, Canada

Yue Lu, China
Nicole Vincent, France
Pong C. Yuen, Hong Kong, China
Wei-Shi Zheng, China
viii Organization

Competition Judges

Mohamed Cheriet, Chair, Canada

Donato Impedovo, Italy
Jay Liebowitz, USA
Jean-Marc Ogier, France
Lihua Yang, China

Exhibitions and Industrial Liaisons

Jun Tan, Chair, China

Alexandre Cornu, Canada
Yufang Tang, China

Publication Chairs

Xiaoyi Jiang, Germany

Muna Khayyat, Canada
Qin Lu, Hong Kong, China

Sponsorship Chairs

Marleah Blom, Canada

Mounîm El Yacoubi, France
Wen Song, China

Financial and Local Arrangements

Louisa Lam, Finance, Canada

Shankang Huang, Liaison, Canada
Connie Chan Kwok, Hong Kong, Canada
Zhongshan Committee, China

Publicity Chairs

Veronique Eglin, France

Soo-Hyung Kim, South Korea
Hongjun Li, China
Li Liu, China
Xiaoqing Lu, China
Jose F Martinez-Trinidad, Mexico
Umapada Pal, India
Alun Preece, UK
Javad Sadri, Canada
Dit-Yan Yeung, Hong Kong, China
 Organization ix

Jia You, HK, China

Bill Wang, Canada
David Zhang, China
Xingming Zhao, China
Jie Zhou, USA
Jun Zhou, Australia

Conference Secretariats and Web Design

Nicola Nobile, Secretariat, Canada

Phoebe Chan, Canada
S. Ling Chan Suen, Canada
Yan Xu, China
 Keynote Speakers

ICPRAI 2020 was proud to host the following keynote speakers:

Hironobu Fujiyoshi Chubu University, Japan “Visualizing Direction of Attention

to Understand AI Decisions”
Seong-Whan Lee Korea University, “Recent Advances and Future
South Korea Prospects of Brain-Computer
Interface”
Cheng-Lin Liu Institute of Automation of “New Frontiers of Document
Chinese Academy of Image Analysis and
Sciences, China Recognition”
Farida Cheriet Polytechnique Montreal, “PRAI in Multimodal Medical
Canada Image Analysis: Applications
and Challenges”
Nicole Vincent and University of Paris, France “Optimal Choices of Features in
Camille Kurtz Image Analysis”
Sheila Lowe American Handwriting “Revealing Human Personality
Analysis Foundation, USA Through Handwriting Analysis”
Bidyut B. Chaudhuri Techno India University, “Improvement of Deep Neural Net
India for Machine Learning”
David Zhang The Chinese University of “Advanced Biometrics: Research
Hong Kong, Hong Kong, and Development”
China
 Sponsors

Sponsored by CENPARMI and other scientific/technological/industrial partners.

xiv Sponsors
 Contents

Handwriting and Text Processing

Gender Detection from Handwritten Documents Using Concept

of Transfer-Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Najla AL-Qawasmeh and Ching Y. Suen

Split and Merge: Component Based Segmentation Network

for Text Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Pan Gao, Qi Wan, and Linlin Shen

Gate-Fusion Transformer for Multimodal Sentiment Analysis . . . . . . . . . . . . 28

Long-Fei Xie and Xu-Yao Zhang

Overview of Mathematical Expression Recognition . . . . . . . . . . . . . . . . . . . 41

Jiashu Huang, Jun Tan, and Ning Bi

Handwritten Mathematical Expression Recognition: A Survey . . . . . . . . . . . 55

Fukeng He, Jun Tan, and Ning Bi

Commodity Classification Based on Multi-modal Jointly Using Image

and Text Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
Yan Xu, Yufang Tang, and Ching Y. Suen

A New DCT-FFT Fusion Based Method for Caption and Scene Text
Classification in Action Video Images . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
Lokesh Nandanwar, Palaiahnakote Shivakumara, Suvojit Manna,
Umapada Pal, Tong Lu, and Michael Blumenstein

A New Method for Detecting Altered Text in Document Images. . . . . . . . . . 93

Lokesh Nandanwar, Palaiahnakote Shivakumara, Umapada Pal,
Tong Lu, Daniel Lopresti, Bhagesh Seraogi, and Bidyut B. Chaudhuri

Hand-Drawn Object Detection for Scoring Wartegg Zeichen Test . . . . . . . . . 109

Nam Tuan Ly, Lili Liu, Ching Y. Suen, and Masaki Nakagawa

Application of Deep Learning for Online Handwritten Mathematical

Expression Recognition: A Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
Danfeng Zhang and Jun Tan

Comparison of Persian Handwritten Digit Recognition in Three Color

Modalities Using Deep Neural Networks . . . . . . . . . . . . . . . . . . . . . . . . . . 125
Abbas Zohrevand, Mahdi Sattari, Javad Sadri, Zahra Imani,
Ching Y. Suen, and Chawki Djeddi
xvi Contents

Application of Deep Learning in Handwritten Mathematical

Expressions Recognition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
Hong Lin and Jun Tan

Automating Stress Detection from Handwritten Documents . . . . . . . . . . . . . 148

Najla AL-Qawasmeh and Muna Khayyat

Recognition of Cursive Caption Text Using Deep Learning -

A Comparative Study on Recognition Units . . . . . . . . . . . . . . . . . . . . . . . . 156
Ali Mirza, Imran Siddiqi, Umar Hayat, Muhammad Atif,
and Syed Ghulam Mustufa

Features and Classifiers

A Hybrid Multiple Classifier System Applied in Life

Insurance Underwriting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171
Chun Lei He, Dave Keirstead, and Ching Y. Suen

Generative Adversarial-Synergetic Networks for Anomaly Detection . . . . . . . 177

Hongjun Li, Chaobo Li, and Ze Zhou

Multi-features Integration for Speech Emotion Recognition . . . . . . . . . . . . . 191

Hongjun Li, Ze Zhou, Xiaohu Sun, and Chaobo Li

Analysis of Multi-class Classification of EEG Signals Using

Deep Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203
Dipayan Das, Tamal Chowdhury, and Umapada Pal

A Deep Object Detection Method for Pineapple Fruit and Flower

Recognition in Cluttered Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
Chen Wang, Jun Zhou, Cheng-yuan Xu, and Xiao Bai

A New Credit Scoring Model Based on Prediction of Signal on Graph . . . . . 228

Zhihua Yang, Qian Zhang, Feng Zhou, and Lihua Yang

Characterizing the Impact of Using Features Extracted

from Pre-trained Models on the Quality of Video Captioning
Sequence-to-Sequence Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
Menatallh Hammad, May Hammad, and Mohamed Elshenawy

Prediction of Subsequent Memory Effects Using Convolutional

Neural Network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251
Jenifer Kalafatovich, Minji Lee, and Seong-Whan Lee

Secure Data Transmission of Smart Home Networks Based

on Information Hiding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264
Haiyu Deng, Lina Yang, Xiaocui Dang, Yuan Yan Tang,
and Patrick Wang
Another random document with
no related content on Scribd:
ionized, then doubly ionized—a sign that the battering becomes
more severe as the heat becomes more intense. (The last stage is
indicated by the disappearance of all visible signs of calcium,
because the ion with two electrons missing has no lines in the
observable part of the spectrum.) The progressive change of other
elements is shown in a similar way. A great advance in this study
was made in 1920 by Professor M. N. Saha, who first applied the
quantitative physical laws which determine the degree of ionization
at any given temperature and pressure. He thereby struck out a new
line in astrophysical research which has been widely developed.
Thus, if we note the place in the stellar sequence where complete
calcium atoms give place to atoms with one electron missing, the
physical theory is able to state the corresponding temperature or
pressure.[16] Saha’s methods have been improved by R. H. Fowler
and E. A. Milne. One important application was to determine the
surface temperatures of the hottest types of stars (12,000°—
25,000°), since alternative methods available for cooler stars are not
satisfactory at these high temperatures. Another rather striking
result was the discovery that the pressure in the star (at the level
surveyed by the spectroscope) is only ¹⁄₁₀₀₀₀th of an atmosphere;
previously it had been assumed on no very definite evidence to be
about the same as that of our own atmosphere.
We commonly use the method of spectrum analysis when we
wish to determine which elements are present in a given mineral on
the earth. It is equally trustworthy in examining the stars since it can
make no difference whether the light we are studying comes from a
body close at hand or has travelled to us for hundreds of years
across space. But one limitation in stellar work must always be
remembered. When the chemist is looking, say, for nitrogen in his
mineral, he takes care to provide the conditions which according to
his experience are necessary for the nitrogen spectrum to show
itself. But in the stars we have to take the conditions as we find
them. If nitrogen does not appear, that is no proof that nitrogen is
absent; it is much more likely that the stellar atmosphere does not
hit off the right conditions for the test. In the spectrum of Sirius the
lines of hydrogen are exceedingly prominent and overwhelm
everything else. We do not infer that Sirius is composed mainly of
hydrogen; we infer instead that its surface is at a temperature near
10,000°, because it can be calculated that that is a temperature
most favourable for a great development of these hydrogen lines. In
the sun the most prominent spectrum is iron. We do not infer that
the sun is unusually rich in iron; we infer that it is at a comparatively
low temperature near 6,000° favourable for the production of the
iron spectrum. At one time it was thought that the prominence of
hydrogen in Sirius and of metallic elements in the sun indicated an
evolution of the elements, hydrogen turning into heavier elements as
the star cools from the Sirian to the solar stage. There is no ground
for interpreting the observations in that way; the fading of the
hydrogen spectrum and the increase of the iron spectrum would
occur in any case as the result of the fall of temperature; and similar
spurious appearances of evolution of elements can be arranged in
the laboratory.
It is rather probable that the chemical elements have much the
same relative abundance in the stars that they have on the earth. All
the evidence is consistent with this view; and for a few of the
commoner elements there is some positive confirmation. But we are
limited to the outside of the star as we are limited to the outside of
the earth in computing the abundance of the elements, so that this
very provisional conclusion should not be pressed unduly.
 Spectral Series
To illustrate further this kind of deduction, let us consider the
spectrum shown in Fig. 9 and see what may be learnt from it. With a
little trouble we can disentangle a beautifully regular series of bright
lines. The marks above will assist you to pick out the first few lines
of the series from the numerous other spectra mixed up with it.
Noticing the diminishing spacing from right to left, you will be able to
see that the series continues to the left for at least fifteen lines
beyond the last one marked, the lines ultimately drawing close
together and forming a ‘head’ to the series. This is the famous
Balmer Series of hydrogen, and having recognized it we identify
hydrogen as one of the elements present in the source of the light.
But that is only the first step, and we can proceed to further
inferences.
Professor Bohr’s theory of the hydrogen atom teaches us that
each line of the series is emitted by an atom in a different state.
These ‘states of excitation’ can be numbered consecutively, starting
from the normal state of the hydrogen atom as No. 1. The light
emitted in the first few states comes into the part of the spectrum
not reproduced here, and the first line in our picture corresponds to
state No. 8. Counting to the left from this you will recognize the
successive lines without much difficulty up to state No. 30. Now the
successive states correspond to more and more swollen atoms, that
is to say, the planet electron[17] makes a wider and wider circuit.
The radius (or more strictly the semi-axis) of its orbit is proportional
to the square of the number of the state, so that the orbit for state
No. 30 is 900 times larger than the orbit for the normal atom No. 1.
The diameter of the orbit in No. 30 is approximately a ten-
thousandth of a millimetre. One inference can be drawn immediately
—the spectrum shown in Fig. 9 was not produced in any terrestrial
laboratory. In the highest vacuum that can be used in terrestrial
spectroscopy the atoms are still too crowded to leave room for an
orbit so large as this. The source must be matter so tenuous that
there is vacant space for the electron to make this wide circuit
without colliding with or suffering interference from other atoms.
Without entering into further detail we can conclude that Fig. 9 is a
spectrum of matter more rarefied than the highest vacuum known
on the earth.[18]
It is interesting to notice that, whereas throughout most of the
picture the lines are shown on a dark background, at the extreme
left the background is bright; the change occurs just at the point
where the Balmer Series comes to an end. This background of light
is also due to hydrogen and it is caused in the following way. The
swollen atoms in state No. 30 or thereabouts are perilously near the
bursting-point, so it is natural that along with them there should be
atoms which have overstepped the limit and burst. They have lost
their planet electrons and are occupied in catching new ones. Just as
energy is required in order to wrench away an electron from an
atom, so there will be superfluous energy to be got rid of when the
atom tames a wild electron. This superfluous energy is radiated and
forms the bright background referred to. Without entering into
technicalities of the theory, we can see that it is appropriate that this
light from the burst atoms should appear in the spectrum
immediately beyond the lines from the most swollen atoms, since
bursting is a sequel to overswelling.
Whilst you have this photograph of the Balmer Series before you
I may take the opportunity of recounting the history of another
famous series. In some of the hottest stars a related series of lines
known as the Pickering Series was discovered in 1896. This is
spaced on precisely the same regular plan, but the lines fall half way
between the lines of the Balmer Series—not exactly half way
because of the gradually diminishing intervals from right to left, but
just where one would naturally interpolate lines in order to double
their number whilst keeping the spacing regular. Unlike the Balmer
Series, the Pickering Series had never been produced in any
laboratory. What element was causing it? The answer seemed
obvious; surely these two related series, one fitting half way
between the other, must belong to different modes of vibration of
the same atom, hydrogen. That seemed to be the only possible
answer at the time; but we have learned more about atoms since
then. We may fairly argue that the ideal simplicity of these two
series indicates that they are produced by an atomic system of the
simplest possible type, viz. an atom with one planet electron; but it
must be remembered that this condition only tells us how the atom
is clothed, not what the atom is. The helium atom (or, for that
matter, the uranium atom) can on occasion masquerade in the
scanty attire of the hydrogen atom. Normal helium has two planet
electrons; but if one of these is lost, it becomes hydrogen-like and
copies the simple hydrogen system on a different scale. It is
significant that the Pickering Series appears only in the very hottest
stars—in conditions likely to cause loss of an electron. The difference
between hydrogen and hydrogen-like helium is firstly the difference
of atomic weight; the helium nucleus is four times as massive. But
this scarcely affects the spectrum because both nuclei are so
massive that they remain almost unshaken by the dancing electron.
Secondly, the helium nucleus has a double electric charge; this is
equivalent to substituting in the vibrating system a controlling spring
of twice the strength. What can be more natural than that the
doubled force of the spring should double the number of lines in the
series without otherwise altering its plan? In this way Professor Bohr
discovered the real origin of the Pickering Series; it is due to ionized
helium, not to hydrogen.[19]
The heavy nucleus, whether of hydrogen or helium, remains
almost unshaken by the atomic vibration—almost, but not quite. At a
later date Professor A. Fowler succeeded in reproducing the
Pickering Series in the laboratory and was able to measure the lines
with much greater accuracy than could be achieved in stellar
spectroscopy; he was then able to show from his measures that the
nucleus is not quite irresponsive. It was a delicate double-star
problem transferred to the interior of the atom; or perhaps a closer
analogy would be the mutual influence of the sun and Jupiter,
because Jupiter, having a thousandth of the mass of the sun,
disturbs it to about the same extent that the light electron disturbs
the hydrogen nucleus. Ionized helium is a faithful copy of the
hydrogen atom (on the altered scale) in everything except the
‘shake’; the shake is less than in hydrogen because the helium
nucleus is still more massive and rock-like. The difference of shake
throws the Pickering Series of helium and the Balmer Series of
hydrogen slightly out of step with respect to one another; and by
measuring this misfit Professor Fowler was able to make a very
accurate determination of the shake and therefore of the mass of
the electron. In this way the mass of the electron is found to be
¹⁄₁₈₄₄th of the mass of the hydrogen nucleus; this agrees well with
the mass found by other methods, and the determination is probably
not inferior in accuracy to any of them.
And so the clue first picked up in stars 300 light years away,
followed in turn by the theoretical and the experimental physicist,
leads in the end to the smallest of all things known.
 The Cloud in Space
Having already considered the densest matter in the universe, we
now turn to consider the rarest.
In spite of great improvements in the art of exhausting vessels
we are still a long way from producing a real vacuum. The atoms in
a vacuum tube before it is exhausted muster a formidable number
containing about twenty digits. High exhaustion means knocking off
five or six noughts at the end of that number; and the most
strenuous efforts to knock off one more nought seem ludicrously
ineffective—a mere nibbling at the huge number that must remain.
Some of the stars are extremely rarefied. Betelgeuse, for
example, has a density about a thousandth that of air. We should
call it a vacuum were it not contrasted with the much greater
vacuosity of surrounding space. Nowadays physicists have no
difficulty in producing a better vacuum than Betelgeuse; but in
earlier times this star would have been regarded as a very creditable
attempt at a vacuum.
The outer parts of a star, and especially the light appendages
such as the solar chromosphere and corona, reach much lower
densities. Also the gaseous nebulae are, as their appearance
suggests, extremely tenuous. When there is space enough to put a
pin’s head between adjacent atoms we can begin to talk about a
‘real vacuum.’ At the centre of the Orion nebula that degree of
rarefaction is probably reached and surpassed.
A nebula has no definite boundary and the density gradually
fades off. There is reason to think that the fading off becomes slow
at great distances. Before we pass entirely out of the sphere of one
nebula we enter the sphere of another, so that there is always some
residual density in interstellar space.
 I believe that, reasoning from the tailing off of the nebulae, we
are in a position to make an estimate of the amount of matter
remaining unaggregated in space. An ordinary region where there is
no observable nebulosity is the highest vacuum existing—within the
limits of the stellar system at least—but there still remains about one
atom in every cubic inch. It depends on our point of view whether
we regard this as an amazing fullness or an amazing emptiness of
space. Perhaps it is the fullness that impresses us most. The atom
can find no place of real solitude within the system of the stars;
wherever it goes it can nod to a colleague not more than an inch
away.
Let us approach the same subject from a different angle.
In the ‘Story of Algol’ I referred to the way in which we measure
the velocity of rotation of the sun. We point the spectroscope first on
one limb of the sun and then on the other. Taking any one of the
dark lines of the spectrum, we find that it has shifted a little
between the two observations. This tells us that the material which
imprinted the line was moving towards or away from us with
different velocities in the two observations. That is what we
expected to find; the rotation of the sun makes solar material move
towards us on one side of the disk and away from us on the other
side. But there are a few dark lines which do not show this change.
They are in just the same position whether we observe them on the
east or on the west of the sun. Clearly these cannot originate on the
sun. They have been imprinted on the light after it left the sun and
before it reached our telescope. We have thus discovered a medium
occurring somewhere between the sun and our telescope; and as
some of the lines are recognized as belonging to oxygen, we can
infer that it is a medium containing oxygen.
This seems to be the beginning of a great discovery, but it ends
in a bathos. It happens that we were already aware of a medium
containing oxygen lying somewhere between our telescope and the
sun. It is a medium essential to our existence. The terrestrial
atmosphere is responsible for the ‘fixed’ lines seen in the sun’s
spectrum.
Just as the spectroscope can tell us that the sun is turning round
(a fact already familiar to us from watching the surface markings),
so it can tell us that certain stars are wandering round an orbit, and
therefore are under the influence of a second star which may or may
not be visible itself. But here again we sometimes find ‘fixed’ lines
which do not change with the others. Therefore somewhere
between the star and the telescope there exists a stationary medium
which imprints these lines on the light. This time it is not the earth’s
atmosphere. The lines belong to two elements, calcium and sodium,
neither of which occur in the atmosphere. Moreover, the calcium is in
a smashed state, having lost one of its electrons, and the conditions
in our atmosphere are not such as would cause this loss. There
seems to be no doubt that the medium containing the sodium and
ionized calcium—and no doubt many other elements which do not
show themselves—is separate from the earth and the star. It is the
‘fullness’ of interstellar space already mentioned. Light has to pass
one atom per cubic inch all the way from the star to the earth, and it
will pass quite enough atoms during its journey of many hundred
billion miles to imprint these dark lines on its spectrum.
At first there was a rival interpretation. It was thought that the
lines were produced in a cloud attached to the star—forming a kind
of aureole round it. The two components travel in orbits round each
other, but their orbital motion need not disturb a diffuse medium
filling and surrounding the combined system. This was a very
reasonable suggestion, but it could be put to the test. The test was
again velocity. Although either component can move periodically to
and fro within the surrounding cloud of calcium and sodium, it is
clear that its average approach to us or recession from us taken over
a long time must agree with that of the calcium and sodium if the
star is not to leave its halo behind. Professor Plaskett with the 72-
inch reflector at the Dominion Observatory in British Columbia
carried out this test. He found that the secular or average rate of
approach of the star[20] was in general quite different from the rate
shown by the fixed calcium or sodium lines. Clearly the material
responsible for the fixed lines could not be an appendage of the star
since it was not keeping pace with it. Plaskett went farther and
showed that whereas the stars themselves had all sorts of individual
velocities, the material of the fixed lines had the same or nearly the
same velocity in all parts of the sky, as though it were one
continuous medium throughout interstellar space. I think there can
be no doubt that this research demonstrates the existence of a
cosmic cloud pervading the stellar system. The fullness of interstellar
space becomes a fact of observation and no longer a theoretical
conjecture.
The system of the stars is floating in an ocean—not merely an
ocean of space, not merely an ocean of ether, but an ocean that is
so far material that one atom or thereabouts occurs in each cubic
inch. It is a placid ocean without much relative motion; currents
probably exist, but they are of a minor character and do not attain
the high speeds commonly possessed by the stars.
Many points of interest arise, but I will only touch on one or two.
Why are the calcium atoms ionized? In the calm of interstellar space
we seem to have passed away from the turmoil which smashed the
calcium atoms in the interior of a star; so at first it seems difficult to
understand why the atoms in the cloud should not be complete.
However, even in the depths of space the breaking-up of the atom
continues; because there is always starlight passing across space,
and some of the light-waves are quite powerful enough to wrench a
first or second electron away from the calcium atom. It is one of the
most curious discoveries of modern physics that when a light-wave is
attenuated by spreading, what it really suffers from is laziness rather
than actual loss of power. What is weakened is not the power but
the probability that it will display the power. A light-wave capable of
bursting an atom still retains the power when it is attenuated a
million-fold by spreading; only it is a million times more sparing in
the exercise of the power. To put it another way, an atom exposed to
the attenuated waves will on the average have to wait a million
times longer before a wave chooses to explode it; but the explosion
when it does occur will be of precisely the same strength however
great the attenuation. This is entirely unlike the behaviour of water-
waves; a wave which is at first strong enough to capsize a boat will,
after spreading, become too weak. It is more like machine-gun fire
which is more likely to miss a given object at greater distance but is
equally destructive if it hits. The property here referred to (the
quantum property) is the deepest mystery of light.
Thus in interstellar space electrons are still being torn from
calcium atoms, only very infrequently. The other side of the question
is the rate of repair, and in this connexion the low density of the
cosmic cloud is the deciding factor. The atom has so few
opportunities for repair. Roving through space the atom meets an
electron only about once a month, and it by no means follows that it
will capture the first one it meets. Consequently very infrequent
smashing will suffice to keep the majority of the atoms ionized. The
smashed state of the atoms inside a star can be compared to the
dilapidation of a house visited by a tornado; the smashed state in
interstellar space is a dilapidation due to ordinary wear and tear
coupled with excessive slackness in making repairs.
A calculation indicates that most of the calcium atoms in
interstellar space have lost two electrons; these atoms do not
interfere with the light and give no visible spectrum. The ‘fixed lines’
are produced by atoms temporarily in a better state of repair with
only one electron missing; they cannot amount at any moment to
more than one-thousandth of the whole number, but even so they
will be sufficiently numerous to produce the observed absorption.
We generally think of interstellar space as excessively cold. It is
quite true that any thermometer placed there would show a
temperature only about 3° above the absolute zero—if it were
capable of registering so low a reading. Compact matter such as a
thermometer, or even matter which from the ordinary standpoint is
regarded as highly diffuse, falls to this low temperature. But the rule
does not apply to matter as rarefied as the interstellar cloud. Its
temperature is governed by other considerations, and it will probably
be not much below the surface-temperature of the hottest stars, say
15,000°. Interstellar space is at the same time excessively cold and
decidedly hot.[21]
 The Sun’s Chromosphere
Once again we shift the scene, and now we are back in the outer
parts of the sun. Fig. 10[22] shows one of the huge prominence
flames which from time to time shoot out of the sun. The flame in
this picture was about 120,000 miles high. It went through great
changes of form and disappeared in not much more than twenty-
four hours. This was rather an exceptional specimen. Smaller flames
occur commonly enough; it seems that the curious black marks in
Fig. 1, often looking like rifts, are really prominences seen in
projection against the still brighter background of the sun. The
flames consist of calcium, hydrogen, and several other elements.
We are concerned not so much with the prominences as with the
layer from which they spring. The ordinary atmosphere of the sun
terminates rather abruptly, but above it there is a deep though very
rarefied layer called the chromosphere consisting of a few selected
elements which are able to float—float, not on the top of the sun’s
atmosphere, but on the sunbeams. The art of riding a sunbeam is
evidently rather difficult, because only a few of the elements have
the necessary skill. The most expert is calcium. The light and nimble
hydrogen atom is fairly good at it, but the ponderous calcium atom
does it best.
The layer of calcium suspended on the sunlight is at least 5,000
miles thick. We can observe it best when the main part of the sun is
hidden by the moon in an eclipse; but the spectroheliograph enables
us to study it to some extent without an eclipse.
Fig. 10. SOLAR PROMINENCE
 Fig. 11. STAR CLUSTER ω CENTAURI

On the whole it is steady and quiescent, although, as the

prominence flames show, it is liable to be blown sky-high by violent
outbursts. The conclusions about the calcium chromosphere that I
am going to describe rest on a series of remarkable researches by
Professor Milne.
How does an atom float on a sunbeam? The possibility depends
on the pressure of light to which we have already referred (p. 26).
The sunlight travelling outwards carries a certain outward
momentum; if the atom absorbs the light it absorbs also the
momentum and so receives a tiny impulse outwards. This impulse
enables it to recover the ground it was losing in falling towards the
sun. The atoms in the chromosphere are kept floating above the sun
like tiny shuttlecocks, dropping a little and then ascending again
from the impulse of the light. Only those atoms which can absorb
large quantities of sunlight in proportion to their weight will be able
to float successfully. We must look rather closely into the mechanism
of absorption of the calcium atom if we are to see why it excels the
other elements.
The ordinary calcium atom has two rather loose electrons in its
attendant system; the chemists express this by saying that it is a
divalent element, the two loose electrons being especially important
in determining the chemical behaviour. Each of these electrons
possesses a mechanism for absorbing light. But under the conditions
prevailing in the chromosphere one of the electrons is broken away,
and the calcium atoms are in the same smashed state that gives rise
to the ‘fixed lines’ in the interstellar cloud. The chromospheric
calcium thus supports itself on what sunlight it can gather in with
the one loose electron remaining. To part with this would be fatal;
the atom would no longer be able to absorb sunlight, and would
drop like a stone. It is true that after two electrons are lost there are
still eighteen remaining; but these are held so tightly that sunlight
has no effect on them and they can only absorb shorter waves which
the sun does not radiate in any quantity. The atom therefore could
only save itself if it restored its main absorbing mechanism by
picking up a passing electron; it has little chance of catching one in
the rarefied chromosphere, so it would probably fall all the way to
the sun’s surface.
There are two ways in which light can be absorbed. In one the
atom absorbs so greedily that it bursts, and the electron scurries off
with the surplus energy. That is the process of ionization which was
shown in Fig. 5. Clearly this cannot be the process of absorption in
the chromosphere because, as we have seen, the atom cannot
afford to lose the electron. In the other method of absorption the
atom is not quite so greedy. It does not burst, but it swells visibly. To
accommodate the extra energy the electron is tossed up into a
higher orbit. This method is called excitation (cf. p. 59). After
remaining in the excited orbit for a little while the electron comes
down again spontaneously. The process has to be repeated 20,000
times a second in order to keep the atom balanced in the
chromosphere.
The point we are leading up to is, Why should calcium be able to
float better than other elements? It has always seemed odd that a
rather heavy element (No. 20 in order of atomic weight) should be
found in these uppermost regions where one would expect only the
lightest atoms. We see now that the special skill demanded is to be
able to toss up an electron 20,000 times a second without ever
making the fatal blunder of dropping it. That is not easy even for an
atom. Calcium[23] scores because it possesses a possible orbit of
excitation only a little way above the normal orbit so that it can
juggle the electron between these two orbits without serious risk.
With most other elements the first available orbit is relatively much
higher; the energy required to reach this orbit is not so very much
less than the energy required to detach the electron altogether; so
that we cannot very well have a continuous source of light capable
of causing the orbit-jumps without sometimes overdoing it and
causing loss of the electron. It is the wide difference between the
energy of excitation and the energy of ionization of calcium which is
so favourable; the sun is very rich in ether-waves capable of causing
the first, and is almost lacking in ether-waves capable of causing the
second.
The average time occupied by each performance is ¹⁄₂₀₀₀₀th of
a second. This is divided into two periods. There is a period during
which the atom is patiently waiting for a light-wave to run into it and
throw up the electron. There is another period during which the
electron revolves steadily in the higher orbit before deciding to come
down again. Professor Milne has shown how to calculate from
observations of the chromosphere the durations of both these
periods. The first period of waiting depends on the strength of the
sun’s radiation. But we focus attention especially on the second
period, which is more interesting because it is a definite property of
the calcium atom, having nothing to do with local circumstances.
Although we measure it for ions in the sun’s chromosphere, the
same result must apply to calcium ions anywhere. Milne’s result is
that an electron tossed into the higher orbit remains there for an
average time of a hundred-millionth of a second before it
spontaneously drops back again. I may add that during this brief
time it makes something like a million revolutions in the upper orbit.
Perhaps this is a piece of information that you were not
particularly burning to know. I do not think it can be called
interesting except to those who make a hobby of atoms. But it does
seem to me interesting that we should have to turn a telescope and
spectroscope on the sun to find out this homely property of a
substance which we handle daily. It is a kind of measurement of
immense importance in physics. The theory of these atomic jumps
comes under the quantum theory which is still the greatest puzzle of
physical science; and it is greatly in need of guidance from
observation on just such a matter as this. We can imagine what a
sensation would be caused if, after a million revolutions round the
sun, a planet made a jump of this kind. How eagerly we should try
to determine the average interval at which such jumps occurred! The
atom is rather like a solar system, and it is not the less interesting
because it is on a smaller scale.
There is no prospect at present of measuring the time of
relaxation of the excited calcium atom in a different way. It has,
however, been found possible to determine the corresponding time
for one or two other kinds of atoms by laboratory experiments. It is
not necessary that the time should be at all closely the same for
different elements; but laboratory measurements for hydrogen also
give the period as a hundred-millionth of a second, so there is no
fault to find with the astronomical determination for calcium.
 The excitation of the calcium atom is performed by light of two
particular wave-lengths, and the atoms in the chromosphere support
themselves by robbing sunlight of these two constituents. It is true
that after a hundred-millionth of a second a relapse comes and the
atom has to disgorge what it has appropriated; but in re-emitting
the light it is as likely to send it inwards as outwards, so that the
outflowing sunlight suffers more loss than it recovers. Consequently,
when we view the sun through this mantle of calcium the spectrum
shows gaps or dark lines at the two wave-lengths concerned. These
lines are denoted by the letters H and K. They are not entirely black,
and it is important to measure the residual light at the centre of the
lines, because we know that it must have an intensity just strong
enough to keep calcium atoms floating under solar gravity; as soon
as the outflowing light is so weakened that it can support no more
atoms it can suffer no further depredations, and so it emerges into
outer space with this limiting intensity. The measurement gives
numerical data for working out the constants of the calcium atom
including the time of relaxation mentioned above.
The atoms at the top of the chromosphere rest on the weakened
light which has passed through the screen below; the full sunlight
would blow them away. Milne has deduced a consequence which
may perhaps have a practical application in the phenomena of
explosion of ‘new stars’ or novae, and in any case is curiously
interesting. Owing to the Doppler effect a moving atom absorbs a
rather different wave-length from a stationary atom; so that if for
any cause an atom moves away from the sun it will support itself on
light which is a little to one side of the deepest absorption. This
light, being more intense than that which provided a balance, will
make the atom recede faster. The atom’s own absorption will thus
gradually draw clear of the absorption of the screen below. Speaking
rather metaphorically, the atom is balanced precariously on the
summit of the absorption line and it is liable to topple off into the full
sunlight on one side. Apparently the speed of the atom should go on
increasing until it has to climb an adjacent absorption line (due
perhaps to some other element); if the line is too intense to be
surmounted the atom will stick part-way up, the velocity remaining
fixed at a particular value. These later inferences may be rather far-
fetched, but at any rate the argument indicates that there is likely to
be an escape of calcium into outer space.
By Milne’s theory we can calculate the whole weight of the sun’s
calcium chromosphere. Its mass is about 300 million tons. One
scarcely expects to meet with such a trifling figure in astronomy. It is
less than the tonnage handled by our English railways each year. I
think that solar observers must feel rather hoaxed when they
consider the labour that they have been induced to spend on this
airy nothing. But science does not despise trifles. And astronomy can
still be instructive even when, for once in a way, it descends to
commonplace numbers.
 The Story of Betelgeuse
This story has not much to do with atoms, and scarcely comes
under the title of these lectures; but we have had occasion to allude
to Betelgeuse as the famous example of a star of great size and low
density, and its history is closely associated with some of the
developments that we are studying.
No star has a disk large enough to be seen with our present
telescopes. We can calculate that a lens or mirror of about 20 feet
aperture would be needed to show traces even of the largest star
disk. Imagine for a moment that we have constructed an instrument
of this order of size. Which would be the most hopeful star to try it
on?
Perhaps Sirius suggests itself first, since it is the brightest star in
the sky. But Sirius has a white-hot surface radiating very intensely,
so that it is not necessary that it should have a wide expanse.
Evidently we should prefer a star which, although bright, has its
surface in a feebly glowing condition; then the apparent brightness
must be due to large area. We need, then, a star which is both red
and bright. Betelgeuse seems best to satisfy this condition. It is the
brighter of the two shoulder-stars of Orion—the only conspicuous
red star in the constellation. There are one or two rivals, including
Antares, which might possibly be preferred; but we cannot go far
wrong in turning our new instrument on Betelgeuse in the hope of
finding the largest or nearly the largest star disk.
You may notice that I have paid no attention to the distances of
these stars. It happens that distance is not relevant. It would be
relevant if we were trying to find the star of greatest actual
dimensions; but here we are considering the star which presents the
largest apparent disk,[24] i.e. covers the largest area of the sky. If
we were at twice our present distance from the sun, we should
receive only one-quarter as much light; but the sun would look half
its present size linearly, and its apparent area would be one-quarter.
Thus the light per unit area of disk is unaltered by distance.
Removing the sun to greater and greater distance its disk will appear
smaller but glowing not less intensely, until it is so far away that the
disk cannot be discriminated.
By spectroscopic examination we know that Betelgeuse has a
surface temperature about 3,000°. A temperature of 3,000° is not
unattainable in the laboratory, and we know partly by experiment
and partly by theory what is the radiating power of a surface in this
state. Thus it is not difficult to compute how large an area of the sky
Betelgeuse must cover in order that the area multiplied by the
radiating power may give the observed brightness of Betelgeuse.
The area turns out to be very small. The apparent size of Betelgeuse
is that of a half-penny fifty miles away. Using a more scientific
measure, the diameter of Betelgeuse predicted by this calculation is
0·051 of a second of arc.
No existing telescope can show so small a disk. Let us consider
briefly how a telescope forms an image—in particular how it
reproduces that detail and contrast of light and darkness which
betrays that we are looking at a disk or a double star and not a blur
emanating from a single point. This optical performance is called
resolving power; it is not primarily a matter of magnification but of
aperture, and the limit of resolution is determined by the size of
aperture of the telescope.
To create a sharply defined image the telescope must not only
bring light where there ought to be light, but it must also bring
darkness where there ought to be darkness. The latter task is the
more difficult. Light-waves tend to spread in all directions, and the
telescope cannot prevent individual wavelets from straying on to
parts of the picture where they have no business. But it has this one
remedy—for every trespassing wavelet it must send a second
wavelet by a slightly longer or shorter route so as to arrive in a
phase opposite to the first wavelet and cancel its effect. This is
where the utility of a wide aperture arises—by affording a wider
difference of route of the individual wavelets, so that those from one
part of the aperture may be retarded relatively to and interfere with
those from another part. A small object-glass can furnish light; it
takes a big object-glass to furnish darkness in the picture.
Now we may ask ourselves whether the ordinary circular
aperture is necessarily the most efficient for giving the wavelets the
required path-differences. Any deviation from a symmetrical shape is
likely to spoil the definition of the image—to produce wings and
fringes. The image will not so closely resemble the object viewed.
But on the other hand we may be able to sharpen up the tell-tale
features. It does not matter how widely the image-pattern may differ
from the object, provided that we can read the significance of the
pattern. If we cannot reproduce a star-disk, let us try whether we
can reproduce something distinctive of a star-disk.
A little reflection shows that we ought to improve matters by
blocking out the middle of the object-glass, and using only the
extreme regions on one side or the other. For these regions the
difference of light-path of the waves is greatest, and they are the
most efficient in furnishing the dark contrast needed to outline the
image properly.
But if the middle of the object-glass is not going to be used, why
go to the expense of manufacturing it? We are led to the idea of
using two widely separated apertures, each involving a
comparatively small lens or mirror. We thus arrive at an instrument
after the pattern of a rangefinder.
This instrument will not show us the disk of a star. If we look
through it the main impression of the star image is very like what we
should have seen with either aperture singly—a ‘spurious disk’
surrounded by diffraction rings. But looking attentively we see that
this image is crossed by dark and bright bands which are produced
by interference between the light-waves coming from the two
apertures. At the centre of the image the waves from the two
apertures arrive crest on crest since they have travelled
symmetrically along equal paths; accordingly there is a bright band.
A very little to one side the asymmetry causes the waves to arrive
crest on trough, so that they cancel one another; here there is a
dark band. The width of the bands decreases as the separation of
the two apertures increases, and for any given separation the actual
width is easily calculated.
Each point of the star’s disk is giving rise to a diffraction image
with a system of bands of this kind, but so long as the disk is small
compared with the finest detail of the diffraction image there is no
appreciable blurring. If we continually increase the separation of the
two apertures and so make the bands narrower, there comes a time
when the bright bands for one part of the disk are falling on the dark
bands for another part of the disk. The band system then becomes
indistinct. It is a matter of mathematical calculation to determine the
resultant effect of summing the band systems for each point of the
disk. It can be shown that for a certain separation of the apertures
the bands will disappear altogether; and beyond this separation the
system should reappear though not attaining its original sharpness.
The complete disappearance occurs when the diameter of the star-
disk is equal to 1⅕ times the width of the bands (from the centre of
one bright band to the next). As already stated, the bandwidth can
be calculated from the known separation of the apertures.
The observation consists in sliding apart the two apertures until
the bands disappear. The diameter of the disk is inferred at once
from their separation when the disappearance occurred. Although
we measure the size of the disk in this way we never see the disk.
We can summarize the principle of the method in the following
way. The image of a point of light seen through a telescope is not a
point but a small diffraction pattern. Hence, if we look at an
extended object, say Mars, the diffraction pattern will blur the fine
detail of the marking on the planet. If, however, we are looking at a
star which is almost a point, it is simpler to invert the idea; the
object, not being an ideal point, will slightly blur the detail of the
diffraction pattern. We shall only perceive the blurring if the
diffraction pattern contains detail fine enough to suffer from it.
Betelgeuse on account of its finite size must theoretically blur a
diffraction pattern; but the ordinary diffraction disk and rings
produced with the largest telescope are too coarse to show this. We
create a diffraction image with finer detail by using two apertures.
Theoretically we can make the detail as fine as we please by
increasing the separation of the two apertures. The method
accordingly consists in widening the separation until the pattern
becomes fine enough to be perceptibly blurred by Betelgeuse. For a
smaller star-disk the same effect of blurring would not be apparent
until the detail had been made still finer by further separation of the
apertures.
This method was devised long ago by Professor Michelson, but it
was only in 1920 that he tried it on a large scale with a great 20-foot
beam across the 100-inch reflector at Mount Wilson Observatory.
After many attempts Pease and Anderson were able to show that the
bright and dark bands for Betelgeuse disappeared when the
apertures were separated 10 feet. The deduced diameter is 0·045 a
second of arc in good enough agreement with the predicted value
(p. 78). Only five or six stars have disks large enough to be
measured with this instrument. It is understood that the construction
of a 50-foot interferometer is contemplated; but even this will be
insufficient for the great majority of the stars. We are fairly confident
that the method of calculation first described gives the correct
diameters of the stars, but confirmation by Michelson’s more direct
method of measurement is always desirable.
To infer the actual size of the star from its apparent diameter, we
must know the distance. Betelgeuse is rather a remote star and its
distance cannot be measured very accurately, but the uncertainty
will not change the general order of magnitude of the results. The
diameter is about 300 million miles. Betelgeuse is large enough to
contain the whole orbit of the earth inside it, perhaps even the orbit
of Mars. Its volume is about fifty million times the volume of the sun.
 There is no direct way of learning the mass of Betelgeuse
because it has no companion near it whose motion it might
influence. We can, however, deduce a mass from the mass-
brightness relation in Fig. 7. This gives the mass equal to 35 x sun.
If the result is right, Betelgeuse is one of the most massive stars—
but, of course, not massive in proportion to its bulk. The mean
density is about one-millionth of the density of water, or not much
more than one-thousandth of the density of air.[25]
There is one way in which we might have inferred that
Betelgeuse is less dense than the sun, even if we had had no
grounds of theory or analogy for estimating its mass. According to
the modern theory of gravitation, a globe of the size of Betelgeuse
and of the same mean density as the sun would have some
remarkable properties:
Firstly, owing to the great intensity of its gravitation, light would
be unable to escape; and any rays shot out would fall back again to
the star by their own weight.
Secondly, the Einstein shift (used to test the density of the
Companion of Sirius) would be so great that the spectrum would be
shifted out of existence.
Thirdly, mass produces a curvature of space, and in this case the
curvature would be so great that space would close up round the
star, leaving us outside—that is to say, nowhere.
Except for the last consideration, it seems rather a pity that the
density of Betelgeuse is so low.

It is now well realized that the stars are a very important adjunct
to the physical laboratory—a sort of high-temperature annex where
the behaviour of matter can be studied under greatly extended
conditions. Being an astronomer, I naturally put the connexion
somewhat differently and regard the physical laboratory as a low-
temperature station attached to the stars. It is the laboratory
conditions which should be counted abnormal. Apart from the
interstellar cloud which is at the moderate temperature of about
15,000°, I suppose that nine-tenths of the matter of the universe is
above 1,000,000°. Under ordinary conditions—you will understand
my use of the word—matter has rather simple properties. But there
are in the universe exceptional regions with temperature not far
removed from the absolute zero, where the physical properties of
matter acquire great complexity; the ions surround themselves with
complete electron systems and become the atoms of terrestrial
experience. Our earth is one of these chilly places and here the
strangest complications can arise. Perhaps strangest of all, some of
these complications can meet together and speculate on the
significance of the whole scheme.
 LECTURE III
THE AGE OF THE STARS

W
E have seen that spatially the scale of man is about midway between the
atom and the star. I am tempted to make a similar comparison as
regards time. The span of the life of a man comes perhaps midway in
scale between the life of an excited atom (p. 74) and the life of a star. For
those who insist on greater accuracy—though I would not like to claim
accuracy for present estimates of the life of a star—I will modify this a little.
As regards mass, man is rather too near to the atom and a stronger claimant
for the midway position would be the hippopotamus. As regards time, man’s
three score years and ten is a little too near to the stars and it would be
better to substitute a butterfly.
There is one serious moral in this fantasy. We shall have to consider
periods of time which appall our imagination. We fear to make such drafts on
eternity. And yet the vastness of the time-scale of stellar evolution is less
remote from the scale of human experience than is the minuteness of the
time-scale of the processes studied in the atom.
Our approach to the ‘age of the stars’ will be devious, and certain
incidental problems will detain us on the way.
 Pulsating Stars
The star δ Cephei is one of the variable stars. Like Algol, its fluctuating
light sends us a message. But the message when it is decoded is not in the
least like the message from Algol.
Let me say at once that experts differ as to the interpretation of the
message of δ Cephei. This is not the place to argue the matter, or to explain
why I think that rival interpretations cannot be accepted. I can only tell you
what is to the best of my belief the correct story. The interpretation which I
follow was suggested by Plummer and Shapley. The latter in particular made
it very convincing, and subsequent developments have, I think, tended to
strengthen it. I would not, however, claim that all doubt is banished.
Algol turned out to be a pair of stars very close together which from time
to time eclipse one another; δ Cephei is a single star which pulsates. It is a
globe which swells and contracts symmetrically with a regular period of 5⅓
days. And as the globe swells and contracts causing great changes of
pressure and temperature in the interior, so the issuing stream of light rises
and falls in intensity and varies also in quality or colour.
There is no question of eclipses; the light signals are not in the form of
‘dots’ and ‘dashes’; and in any case the change of colour shows that there is
a real change in the physical condition of the source of the light. But at first
explanations always assumed that two stars were concerned, and aimed at
connecting the physical changes with an orbital motion. For instance, it was
suggested that the principal star in going round its orbit brushed through a
resisting medium which heated its front surface; thus the light of the star
varied according as the heated front surface or cooler rear surface was
presented towards us. The orbital explanation has now collapsed because it is
found that there is literally no room for two stars. The supposed orbit had
been worked out in the usual way from spectroscopic measurements of
velocity of approach and recession; later we began to learn more about the
true size of stars, first by calculation, and afterwards (for a few stars) by
direct measurement. It turned out that the star was big and the orbit small;
and the second star if it existed would have to be placed inside the principal
star. This overlapping of the stars is a reductio ad absurdum of the binary
hypothesis, and some other explanation must be found.
 What had been taken to be the approach and recession of the star as a
whole was really the approach and recession of the surface as it heaved up
and down with the pulsation. The stars which vary like δ Cephei are diffuse
stars enormously larger than the sun, and the total displacement measured
amounts to only a fraction of the star’s radius. There is therefore no need to
assume a bodily displacement of the star (orbital motion); the measures
follow the oscillation of that part of the star’s surface presented towards us.
The decision that δ Cephei is a single star and not double has one
immediate consequence. It means that the period of 5⅓ days is intrinsic in
the star and is therefore one of the clues to its physical condition. It is a free
period, not a forced period. It is important to appreciate the significance of
this. The number of sunspots fluctuates from a maximum to minimum and
back to maximum in a period of about 11½ years; although we do not yet
understand the reason for this fluctuation, we realize that this period is
something characteristic of the sun in its present state and would change if
any notable change happened to the sun. At one time, however, there was
some speculation as to whether the fluctuation of the sunspots might not be
caused by the revolution of the planet Jupiter, which has a period not so very
different; if that explanation had been tenable the 11½-year period would
have been something forced on the sun from without and would teach us
nothing as to the properties of the sun itself. Having convinced ourselves that
the light-period of δ Cephei is a free period of a single star, belonging to it in
the same way that a particular note belongs to a tuning-fork, we can accept it
as a valuable indicator of the constancy (or otherwise) of the star’s physical
condition.
In stellar astronomy we usually feel very happy if we can determine our
data—parallax, radius, mass, absolute brightness, &c.—to within 5 per cent.;
but the measurement of a period offers chances of far superior accuracy. I
believe that the most accurately known quantity in the whole of science
(excluding pure mathematics) is the moon’s mean period, which is commonly
given to twelve significant figures. The period of δ Cephei can be found to six
significant figures at least. By fastening an observable period to the intrinsic
conditions of a star we have secured an indicator sensitive enough to show
extremely small changes. You will now guess why I am approaching ‘the age
of the stars’ through the Cepheid variables. Up to the present they are the
only stars known to carry a sensitive indicator, by which we might hope to
test the rate of evolutionary change. We believe that δ Cephei like other stars
has condensed out of a nebula, and that the condensation and contraction
are still continuing. No one would expect to detect the contraction by our