Advances in Growth Curve and

Structural Equation Modeling Topics

from the Indian Statistical Institute
Proceedings 2017 Ratan Dasgupta
Visit to download the full and correct content document:
https://textbookfull.com/product/advances-in-growth-curve-and-structural-equation-mo
deling-topics-from-the-indian-statistical-institute-proceedings-2017-ratan-dasgupta/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...

Advances in Growth Curve and Structural Equation

Modeling Topics from the Indian Statistical Institute
on the 125th Birth Anniversary of PC Mahalanobis Ratan
Dasgupta
https://textbookfull.com/product/advances-in-growth-curve-and-
structural-equation-modeling-topics-from-the-indian-statistical-
institute-on-the-125th-birth-anniversary-of-pc-mahalanobis-ratan-
dasgupta/

Multilevel Structural Equation Modeling Bruno Castanho

Silva

https://textbookfull.com/product/multilevel-structural-equation-
modeling-bruno-castanho-silva/

Proceedings of SECON'19: Structural Engineering and

Construction Management Kaustubh Dasgupta

https://textbookfull.com/product/proceedings-of-
secon19-structural-engineering-and-construction-management-
kaustubh-dasgupta/

Proceedings of SECON 2020: Structural Engineering and

Construction Management Kaustubh Dasgupta

https://textbookfull.com/product/proceedings-of-
secon-2020-structural-engineering-and-construction-management-
kaustubh-dasgupta/
Multiple Regression and Beyond An Introduction to
Multiple Regression and Structural Equation Modeling
Timothy Z. Keith

https://textbookfull.com/product/multiple-regression-and-beyond-
an-introduction-to-multiple-regression-and-structural-equation-
modeling-timothy-z-keith/

Special Topics in Structural Dynamics, Volume 5:

Proceedings of the 36th IMAC, A Conference and
Exposition on Structural Dynamics 2018 Nikolaos
Dervilis
https://textbookfull.com/product/special-topics-in-structural-
dynamics-volume-5-proceedings-of-the-36th-imac-a-conference-and-
exposition-on-structural-dynamics-2018-nikolaos-dervilis/

Corporate Governance and Contingency Theory: A

Structural Equation Modeling Approach and Accounting
Risk Implications 1st Edition Abdul Ghofar

https://textbookfull.com/product/corporate-governance-and-
contingency-theory-a-structural-equation-modeling-approach-and-
accounting-risk-implications-1st-edition-abdul-ghofar/

A Primer on Partial Least Squares Structural Equation

Modeling PLS SEM 2nd Edition Joseph F. Hair Jr

https://textbookfull.com/product/a-primer-on-partial-least-
squares-structural-equation-modeling-pls-sem-2nd-edition-joseph-
f-hair-jr/

Special Topics in Structural Dynamics & Experimental

Techniques, Volume 5: Proceedings of the 37th IMAC, A
Conference and Exposition on Structural Dynamics 2019
Nikolaos Dervilis
https://textbookfull.com/product/special-topics-in-structural-
dynamics-experimental-techniques-volume-5-proceedings-of-
the-37th-imac-a-conference-and-exposition-on-structural-
Ratan Dasgupta Editor

Advances in
Growth Curve and
Structural Equation
Modeling
Topics from the Indian Statistical
Institute—Proceedings 2017
Advances in Growth Curve and Structural Equation
Modeling
Ratan Dasgupta
Editor

Advances in Growth Curve

and Structural Equation
Modeling
Topics from the Indian Statistical
Institute—Proceedings 2017

123
Editor
Ratan Dasgupta
Theoretical Statistics
and Mathematics Unit
Indian Statistical Institute
Kolkata, West Bengal, India

ISBN 978-981-13-0979-3 ISBN 978-981-13-0980-9 (eBook)

https://doi.org/10.1007/978-981-13-0980-9

Library of Congress Control Number: 2018943729

© Springer Nature Singapore Pte Ltd. 2018

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, express 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 Singapore Pte Ltd.
The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721,
Singapore
Preface

A growth curve is an empirical model of the evolution of certain characteristic of

interest over time. Growth curve models are widely studied in different branches of
science. The present volume on growth curve model (GCM) is a culmination
of the talks given at the workshop on the topic held during 23–24 February 2017 at
the Indian Statistical Institute, Giridih, where growth experiments are going on in
the farmland.
Professor Jayanta K. Ghosh, an eminent academician and a former director of
the Indian Statistical Institute, passed away on September 30, 2017. He was one
of the inspiring forces in conducting growth experiments in Giridih over years. The
present volume is dedicated to his memory.
This workshop proceeding ‘Advances in Growth Curve and Structural Equation
Modeling: Proceedings 2017—Topics from the Indian Statistical Institute’ presents
some ideas about the research works, both theoretical and applied, on growth curve
models, which are going on by the scientists of Indian Statistical Institute in dif-
ferent branches of science over years. Growth curve models are used with great
effectiveness in many areas of developmental research. I am thankful to the readers
as the previous three volumes on GCM—Advances in Growth Curve Models:
Topics from the Indian Statistical Institute, 2013, Growth Curve and Structural
Equation Modeling: Topics from the Indian Statistical Institute, 2015, and Growth
Curve Models and Applications, Indian Statistical Institute, Giridih, India, March
28–29, 2016 (2017)—are well accepted by the scientific community.
We invited for contribution to this workshop proceeding and further invited the
participants of the workshop to submit more than one paper, if possible for the
proceedings. All the papers were peer-reviewed. The result is a compilation of 11
research papers presented in this volume.
The endeavor will be considered successful, if this can give some idea about
solving theoretical and practical problems in this broad area of GCM in which many
researchers in different branches of science are interested in.

Kolkata, India Ratan Dasgupta

October 2017

v
Contents

Longitudinal Studies on Mathematical Aptitude and Intelligence

Quotient of North Eastern Tribes in Tripura . . . . . . . . . . . . . . . . . . . . . 1
Ratan Dasgupta
Distribution of Mangroves and Soil Parameters in the Lothian
Island of Sundarbans, India—A GIS Approach . . . . . . . . . . . . . . . . . . . 25
Hema Gupta (Joshi) and Monoranjan Ghose
Characterization of Extended Uniform Distribution and
Its Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
Ratan Dasgupta
Do We Share the Same Destiny? An Empirical Study
on the Convergence of Growth Among Indian States . . . . . . . . . . . . . . . 59
Sattwik Santra and Samarjit Das
Cross-Sectional Studies on Mathematical Aptitude
and Intelligent Quotient in North Eastern Tribes . . . . . . . . . . . . . . . . . . 71
Ratan Dasgupta
Growth Models for Repeated Measurement Mixture Experiments:
Optimal Designs for Parameter Estimation and Growth
Prediction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Manisha Pal, Nripes K. Mandal and Bikas K. Sinha
Tuber Crop Growth Model, Performance Rate, and Some
Characterization Theorems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
Ratan Dasgupta
An Overview of Growth and Malnutrition Among 5–18-Year
Children in India . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
Susmita Bharati, Manoranjan Pal and Premananda Bharati

vii
viii Contents

Retrospective Longitudinal Studies on Lifestyle and Social

Environment in North-Eastern Tribes . . . . . . . . . . . . . . . . . . . . . . . . . . 123
Ratan Dasgupta
Microevolution Among the Yanadi Tribe: Population Structure
and Sex Differences in Anthropometric and Dermatoglyphic
Characters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
T. S. Vasulu
A Statistical Study of Platelet Size Decomposition and Related
Growth Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175
Ratan Dasgupta
Some Snapshots from the Workshop on Growth Curve Model,
held during 23–24 February 2017 at the Indian Statistical Institute,
Giridih . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
Editor and Contributors

About the Editor

Prof. Ratan Dasgupta is Senior Professor of Statistics at the Indian Statistical

Institute, Kolkata. His research interests include nonparametric statistics, rates of
convergence in central limit theorem, the application of statistics to industrial
quality control, sociology, education, agriculture, biology, physics, and other nat-
ural sciences. He has been at the forefront of promoting the theory and the appli-
cations of growth curve modeling, and his knowledge, expertise and extensive
publication record on the topic, together with his outstanding theoretical skills make
him uniquely qualified to edit this volume.

Contributors

Premananda Bharati Biological Anthropology Unit, Indian Statistical Institute,

Kolkata, India
Susmita Bharati Sociological Research Unit, Indian Statistical Institute, Kolkata,
India
Samarjit Das Economic Research Unit, Indian Statistical Institute, Kolkata, India
Ratan Dasgupta Theoretical Statistics and Mathematics Unit, Indian Statistical
Institute, Kolkata, West Bengal, India
Monoranjan Ghose Agricultural and Ecological Research Unit, Indian Statistical
Institute, Kolkata, India
Hema Gupta (Joshi) Agricultural and Ecological Research Unit, Indian Statistical
Institute, Kolkata, India; Department of Botany, Visva-Bharati, Bolpur,
Santiniketan, West Bengal, India
Nripes K. Mandal Calcutta University, Kolkata, India

ix
x Editor and Contributors

Manisha Pal Calcutta University, Kolkata, India

Manoranjan Pal Economic Research Unit, Indian Statistical Institute, Kolkata,
India
Sattwik Santra Centre for Training and Research in Public Finance and Policy,
Centre for Studies in Social Sciences, Kolkata, India
Bikas K. Sinha Indian Statistical Institute, Kolkata, India
T. S. Vasulu Indian Statistical Institute, Kolkata, India
Longitudinal Studies on Mathematical
Aptitude and Intelligence Quotient
of North Eastern Tribes in Tripura

Ratan Dasgupta

Abstract Longitudinal studies are conducted on mathematical aptitude and intelli-

gence quotient on North Eastern tribes in Tripura over successive interviews in a time
span of more than 3 years, viz., 20 September 2011–28 November 2014. Analyzed
longitudinal data indicate that both mathematical aptitude and intelligent quotient
scores exhibit fluctuations over time and have upward trend immediately after first
interaction with the interviewer, before stabilizing at a level slightly below the peak
value of scores. Average level of mathematical aptitude is low, although the level
of intelligence quotient score is comparatively high. Growth curves under different
setups are estimated to infer about the status of tribal education and lifestyle. The
score status is seen to be improving over time, although associated with mild fluctu-
ations. Proliferation rates of different scores are estimated under different assump-
tions. In general, the proliferation rates reach stability towards the end of curves
for large values of time. Postulating a simple model of association in scores over
time based on martingales, we examine the fluctuation of scores. Excessive devia-
tion results for martingales are derived. Under certain conditions on the martingale
{Mi : 1 ≤ i ≤ n}, the excessive deviation P(max1≤i≤n |Mi | ≥ λn 1/2 ) is seen to be
λ2
O(e− 2 (1+o(1)) ); λ → ∞. This is similar to the tail probability of normal distribu-
tion. Deviation of observations from response curve may be compared with normal
deviate to detect the presence of assignable causes.

Keywords Lifestyle status · Mathematical aptitude · IQ · Proliferation rate ·

Martingale · Excessive deviation · Kokborok

MS Subject Classification: 62P25 · 60G20

R. Dasgupta (B)
Theoretical Statistics and Mathematics Unit, Indian Statistical Institute,
203 B.T. Road, Kolkata 700108, India
e-mail: rdgupta@isical.ac.in; ratandasgupta@gmail.com

© Springer Nature Singapore Pte Ltd. 2018 1

R. Dasgupta (ed.), Advances in Growth Curve and Structural Equation Modeling,
https://doi.org/10.1007/978-981-13-0980-9_1
2 R. Dasgupta

1 Introduction and Sketch of the Work

Studies on mathematical aptitude and intelligence quotient in different ethnic groups

are of interest. These scores can successfully be utilized to predict the potential
of individuals in different fields of activities at a later stage of life. For a group
of people, high level of the scores may be interpreted as an indication of educa-
tion and general development. Studies conducted in a vulnerable section of soci-
ety indicated that exploitation and education are inversely related, e.g., see p-321
of: http://dspace.vidyanidhi.org.in:8080/dspace/bitstream/2009/4402/10/OSM-
1993-041-9.pdf.
The whole of the North East has been categorized as industrially backward. See,
e.g., http://planningcommission.nic.in/reports/publications/pub_neregion.pdf.
In terms of development, many of the North Eastern states lags behind the rest
of the country and are not able to come to the forefront of education, see e.g.,
http://timesofindia.indiatimes.com/india/7-of-8-Northeast-states-lag-behind-
average-India-income/articleshow/5187537.cms.
We study the above two traits, viz., Mathematical aptitude and IQ in tribal indi-
viduals in northeastern part of Tripura state with 15 simple basic questions related
to mathematical ability and 7 questions on general knowledge and reasoning. In the
beginning of the investigation, the questions in interviews were asked and explained
verbally by the interviewer. Some answers were provided by tribal individuals in
writing and others were recorded by interviewer. In the later stage of investiga-
tion, written questionnaires were provided, the answers were given by the tribal
individuals in between the blank spaces of printed questions. Total score and the
component scores on IQ and Math aptitude are standardized in the range [0, 100]
before plotting.
Analyzed data indicate an upward trend immediately after the first interaction
of individuals with the interviewer, before stabilizing at a level slightly below the
peak. This may be due to the fact that increase in score can be achieved, in general,
by coaching people in the way of answering/explaining items on intelligent test.
Coaching can create exceptional increase of a temporary nature. Also, any increase
in scores tends to level out after several coaching sessions and ultimately that fades
with time, see Duminy et al. (1991).
A similar phenomenon may happen when the tribal individuals are discussing
lifestyle with interviewer in the course of recording answers to such questions on
lifestyle and IQ posed to them for the first time. Here, coaching can be compared with
interaction of tribal individuals with the interviewer when subject matter is explained
in a lucid manner. Impact of this fruitful interaction is reflected in the scores of the
next interview.
Language of instruction for many schools in Tripura is Bengali, while the mother
tongue of a large segment of students is Kok Borok in tribal belt. Linguistic handicaps
are experienced where medium of school instructions is not the mother tongue. Unless
children can acquire complete facility in the language at school, inevitably they are
retarded in reasoning as well as attainment. Often they can learn to speak the language
Longitudinal Studies on Mathematical Aptitude and Intelligence Quotient … 3

fairly accurately, but they do not think naturally in it; instead they tend to revert to
the less efficient but more deeply rooted mother tongue.
The individuals are interviewed at different places of congress like students in
different schools, marketplaces, church, and villages after stratification. Selection of
individuals for interview is made by simple random sampling without replacement
at final stage. Non-response of a unit is remedied by selecting an adjacent available
unit.
In this pilot survey, data are collected in the way of personal interviews, admin-
istering tests and filling up the questionnaire by selected individuals. Sometimes
resistance to fill up the questionnaire was faced from skeptic elements, motif and
ultimate goal of such studies were questioned and monetary angle out of the results
in conducted survey was prodded into, while working in some interior parts of the
state. It appears that affinity of inhabitants of interior hilly region and plane region
where modern facilities are available is slowly improving over time with progress of
education and lifestyle.
In the next section, we present the results obtained from the data analyzed. The
error components in the scores may not be symmetric in all cases.
The individuals may not be highly educated in general, but their receptivity of
to new concepts is commendable. Figures based on scores provide insight into the
lifestyle and educational status of individuals interviewed.
Fluctuations in scores are modeled by martingale differences. Excessive deviation
results for martingale are obtained.
In Sect. 2, we present the results in brief. These are further elaborated in figures
with legends and associated explanation given in detail therein. Section 3 discusses
the results obtained. Fluctuations of observations around the response curves are
explained by martingale theory. While estimating the response curve by nonparamet-
ric regression, assume that the differences of the existing data point from regression
curve behave like martingale differences as new data are gathered from recent inter-
view to modify the regression, and those successive martingale differences constitute
a martingale. Maximal fluctuation of the martingales is of interest.
Excessive deviation results for maximal fluctuations are proved in the Appendix.
Under assumptions on moment bounds, those are not so stringent on martingale dif-
ferences, and deviation probabilities of maximal fluctuations in martingales are seen
to decay with order higher than any polynomial power of the boundary value. Since
the total score and the component scores on IQ and Math aptitude are standardized
in the range [0, 100], we may assume that the martingale differences are bounded.
In such a situation, for the martingales {Mi : 1 ≤ i ≤ n}, we compute the exces-
λ2
sive deviation probability as P(max1≤i≤n |Mi | ≥ λn 1/2 ) = O(e− 2 (1+o(1)) ); λ →
∞, which is similar to the tail probability of normal distribution. High oscillations
of observations from the response curve may be assessed in terms of normal deviate
to identify assignable causes.
In general, the level of IQ is seen to be higher than that of scores in mathematical
aptitude. A large segment of the individuals under the scope of study understood the
4 R. Dasgupta

Fig. 1 Longitudinal score (Math plus IQ) of 47 tribal individuals in Tripura. Longitudinal growth
curves of total scores for 47 tribal individuals in Mathematics and IQ after scaling are shown in
Fig. 1, the maximum possible score is scaled to 100. The number of Math questions is 15, and the
number of IQ questions is 7 with a total of 22 questions administered in interviews. Indications of
upward trend in second interview scores are possibly due to fruitful interaction with the individuals
at the first interview date, on topics at introductory level, before these tests were administered.
Elaboration of the subject matter affects different individuals over a considerable period of time,
and this phenomenon is apparent in observed peaks. Too many lines make the picture congested.
Some individuals appeared only twice, which results in two interview scores over time

aim of the tests and were cooperative with the interviewer on successive occasions
of test administration.

2 Results

The survey conducted over a time span of about 3 years revealed that IQ test score in
percentage is at a higher level than the Math score percentage. Data analyzed under
the assumption that error component is symmetrically distributed and is consistent
with other nonparametric methods to estimate the growth curve for total score of
Math plus IQ, and Math score alone. Symmetry of error component may not hold for
IQ score, expressed in percentage. An upward trend is seen on scores after interaction
with the interviewer, indicating receptivity of individuals towards improvement of
knowledge. Error component in IQ score may not be symmetric. From the growth
curve of Math aptitude and IQ scores, it appears that the education status of tribal
individuals is improving over time. Some of the scores seem to attain stability over
time towards the far end. Different features of collected data and growth curve related
results are further explained in the Figures.
Longitudinal Studies on Mathematical Aptitude and Intelligence Quotient … 5

Fig. 2 Longitudinal score (IQ) of 47 tribal individuals. Questions on Math and IQ are separated and
scores are also separated into two components and expressed in percentage. The first component
IQ score in percentage for longitudinal study is shown in Fig. 2. Patterns in these growth curves
are similar to that of Fig. 1, the level of IQ scores has relatively higher elevation than total score
expressed in percentage

Fig. 3 Longitudinal score (Math) of 47 tribal individuals. The second component, Math score in
longitudinal study is shown in Fig. 3. Elevation of scores is a bit lower that earlier two graphs, and
pattern of the graph is the same with former two graphs. Overall performance in Mathematics seems
to be poor compared to that in IQ

3 Discussions

Investigation on the longitudinal aspect of scores in Mathematical aptitude and IQ on

a segment of tribal population revealed that although the general level of education
is low, improvement over time is observed especially after first exposure to such
aptitude tests and interaction with the interviewer. In general, the level of IQ is
higher than that of scores in mathematical aptitude. Receptivity of the individuals
to new concepts and ideas is commendable. The response curve for the two traits
under investigation may be examined under the mild assumption that the errors are
martingale. Excessive deviation results for maximal fluctuations of martingales are
derived under different moment assumptions. The results are similar to the normal
tail probabilities. Probability of high fluctuation of observations from the response
6 R. Dasgupta

Fig. 4 Band of growth curves (Math plus IQ) of 47 individuals. The band that contains all 47
longitudinal curves on total score is shown by joining the extreme points with straight lines. For
each value of time x, the line of y coordinate has shortest length within the band containing all
individual curves. The curve joining mid points of the upper and lower band is also shown in
Fig. 4. The curve in the middle represents an estimate of overall growth in total scores under the
assumption that error components, i.e., deviation of scores from response curve for each time point
t have symmetric distribution. Mid-range is sometimes more efficient than the mean for estimating
central tendency in a number of distributions with limited range, see, e.g., Rider (1957)

Fig. 5 Lowess on mid points of growth band (Math + IQ) of 47 individuals. A nonparametric
growth curve of total score by lowess regression on mid points of the band with f = 2/3 is shown
in Fig. 5. The curve has initial upward trend for a while, it goes down and reaches a minimum
when initial boost from interaction with interviewer possibly erodes. Then the curve shows an
upward trend gradually over time. We shall see later that the pattern of the growth curve remains
the same, if the median of the scores on days on recording is considered instead of midpoints of the
band considered in lowess regression. This is in concordance with the assumption that errors have
symmetric distribution for total score

curve may then be compared with normal tail probability to investigate for assignable
causes.
Longitudinal Studies on Mathematical Aptitude and Intelligence Quotient … 7

Fig. 6 Band of growth

curves (IQ) of 47 individuals.
A similar band containing of
47 longitudinal curves on IQ
score is shown in Fig. 6. Here
again, the curve in the
middle represents an
estimate of overall growth of
IQ score under the
assumption that errors have
symmetric distribution

Fig. 7 Lowess on mid points of growth band (IQ) of 47 individuals. A nonparametric growth curve
of IQ score by lowess regression on midpoints of the band with f = 2/3 is shown in Fig. 7. The
curve has initial downward trend and reaches a minimum, then the curve gradually exhibits an
upward trend over time. We shall see later in Fig. 26 that the pattern of the graph changes when
the assumption of symmetric error is not made, indicating that the assumption of symmetric error
component for IQ scores may not hold

Fig. 8 Band of growth curves (Math) of 47 individuals. The band containing 47 longitudinal curves
on mathematics score is shown in Fig. 8. The middle curve joining the midpoints of the band provides
an estimate of growth curve in mathematics score
8 R. Dasgupta

Fig. 9 Lowess on midpoints of growth band (Math) of 47 individuals. A nonparametric growth

curve of mathematics score by lowess regression on midpoints of the band with f = 2/3 is shown
in Fig. 9. The curve has an initial upward trend for a while, then it goes down and reaches a
minimum when initial boost from interaction with interviewers possibly erodes; in later part, the
curve gradually shows an upward trend over time
0.0008
Proliferation rate of score/day
0.0006
0.0004
0.0002
0.0
-0.0002

0 200 400 600 800 1000

Time (day)

Fig. 10 Proliferation rate of total score: wt. exp(−x); spline. We compute the proliferation rate
d
based on the growth curve shown in Fig. 5. Proliferation rate dt log y = 1y dy
dt is a scaled version of
velocity dy
dt . This measure is independent of the choice of unit used in measuring y. For total score
y with growth curve computed in Fig. 5 by lowess regression, the proliferation rate is obtained in
Fig. 10. The curve has an initial downward trend up to day 190, and then the curve rises gradually,
with fast growth towards the end. Computation is based on a technique proposed in Dasgupta
(2015), with exponentially decaying weights attached to empirical slopes computed from data pairs
at different time points with respect to a fixed time point t of interest. More weights are given to
data points near the time t of derivative computation, and less weights to distant time points from t.
Weighted mean of these empirical slopes at derivative stage and smooth.spline with spar = 0.01 at
smoothing stage in SPlus provide proliferation rate at time point t, when divided by y. Instead of
considering median or trimmed mean that produces too smooth rates especially towards end of the
range, we take weighted average of empirical slopes; with less weight assigned to slopes computed
from distant points
Longitudinal Studies on Mathematical Aptitude and Intelligence Quotient … 9

0.0010
0.0005
Proliferation rate of score/day
0.0
-0.0005
-0.0010

0 200 400 600 800 1000

Time (day)

d
Fig. 11 Proliferation rate of IQ score: wt. exp(−x); spline. Proliferation rate dt log y over 11 time
points for IQ score y is computed as in Fig. 10, with exponentially decaying weights attached to
empirical slopes computed from data pairs at different time points with respect to a fixed time
point t of interest. Weighted mean of the empirical slopes is considered at derivative stage of
computation and the package smooth.spline with spar = 0.01 at smoothing stage in SPlus provides
the proliferation rate at the time point t. A consistent sharp rise in the beginning and then slightly
slow rise of rate towards the end is seen in Fig. 11. However, we shall see later that the pattern
changes when we consider median instead of mid-values of the band to estimate central tendency.
This is because in Fig. 7, the basic input shown for computing proliferation rate here changes to a
different pattern when the assumption of symmetric error for IQ is dropped
0.0010
Proliferation rate of score/day
0.0005
0.0
-0.0005
-0.0010

0 200 400 600 800 1000

Time (day)

d
Fig. 12 Proliferation rate of math score: wt. exp(−x); spline. Proliferation rate dt log y over
11 time points for Math score y is computed as in Fig. 10, with exponentially decaying weights
attached to empirical slopes computed from data pairs at different time points with respect to a fixed
time point t of interest. Weighted mean of the empirical slopes is considered at derivative stage
of computation and the package smooth.spline with spar = 0.0001 at smoothing stage in SPlus
provide the proliferation rate at the time point t. A downward trend in the beginning and then a
typical sharp rise towards middle is observed in Fig. 12; the rate slows down towards the end and
approaches stability over time
10 R. Dasgupta

Fig. 13 Longitudinal score (Math plus IQ) of 19 individuals with more than two recordings.
Analysis based on longitudinal data with many recordings per individuals has accurate assertions.
We now consider only those cases where more observations are available per individuals, as there are
some longitudinal growth curves with only two recordings. When tracked over time, 19 individuals
had more than two recordings. Total scores of Math plus IQ for these individuals are shown in
Fig. 13. Figure 13 is less messy compared to Fig. 1. Upward trend of scores in general are seen after
the first interview

Fig. 14 Longitudinal score (IQ) of 18 individuals with more than two recordings. Scores on IQ for
18 individuals having more than two recordings and nonzero score are shown in Fig. 14. The scores
are of higher elevation compared to that for total score of previous picture. Upward trend of scores
is more prominent in the second interview, compared to the previous figure

Appendix

To study the deviation of the observed growth from nonparametric response curve
derived under mild assumption and to have a bound on maximum fluctuation of
deviations, consider these errors to be martingale differences. An assignable cause
may be looked into, if the fluctuations are of different type other than specified by
Longitudinal Studies on Mathematical Aptitude and Intelligence Quotient … 11

Fig. 15 Longitudinal score (Math) of 19 individuals with more than two recordings. Scores on
Math for 19 individuals with more than two recordings are shown in Fig. 15. The scores are of
lower elevation compared to those for total score and IQ scores of previous pictures. Upward trend
of scores is prominent in general immediately after the first interview

Fig. 16 Band of growth curves (Math + IQ) of 19 Individuals with >2 recordings. The band
containing 19 longitudinal curves on total score is shown. The curve joining mid points of the upper
and lower band is also shown in Fig. 16. The curve in the middle represents an estimate of overall
growth in total scores under the assumption that errors are symmetric. With less number of curves,
the picture is less messy. An individual scoring zero in IQ, scored a positive mark in Math. Thus,
the number of individuals with nonzero total score is 19 here

Fig. 17 Lowess on mid points of growth band (Math + IQ) of 19 individuals with >2 recordings.
A nonparametric growth curve of total score by lowess regression on mid points of the band with
f = 2/3 is shown in Fig. 17. The curve goes down over time towards stability
12 R. Dasgupta

Fig. 18 Band of growth curves (IQ) of 19 individuals with >2 recordings. A similar band containing
of 18 longitudinal curves on IQ score is shown in Fig. 18. Here again, the curve in the middle
represents an estimate of overall growth of IQ score when errors have symmetric distribution. The
pattern of growth is clearer with less number of individuals

Fig. 19 Lowess on mid points of growth band (IQ) of 18 Individuals with >2 recordings. A
nonparametric growth curve of IQ score by lowess regression on midpoints of the band with f = 2/3
is shown in Fig. 19. The curve has slight upward trend initially after a drop, and gradually lowers
over time. This feature of the graph may be due to the assumption made on symmetric error

Fig. 20 Band of growth curves (Math) of 19 individuals with >2 recordings. The band containing
19 longitudinal curves on Math score is shown in Fig. 20. The curve in the middle represents an
estimate of overall growth of Math score. The oscillatory pattern of growth has slight upward trend.
This becomes clear with less number of individuals
Longitudinal Studies on Mathematical Aptitude and Intelligence Quotient … 13

Fig. 21 Lowess on midpoints of growth band (Math) of 19 individuals with >2 recordings. A
nonparametric growth curve of Math score by lowess regression on midpoints of the band with
f = 2/3 is shown in Fig. 21. The curve indicates an upward trend, after slowing down slightly at
the initial stage
-0.001
Proliferation rate of score/day
-0.002
-0.003
-0.004

0 200 400 600 800 1000

Time (day)

Fig. 22 Proliferation rate of 19 persons total score: wt. exp(−x); spline. We compute the prolifer-
d
ation rate based on the growth curve shown in Fig. 17. Proliferation rate dt log y over 9 time points
for total score y is computed based on a technique proposed in Dasgupta (2015), with exponentially
decaying weights attached to empirical slopes computed from data pairs at different time points
with respect to a fixed time point t of interest. Weighted mean of the empirical slopes is considered
at derivative stage of computation and the package smooth.spline with spar = 0.01 at smoothing
stage in SPlus provides the proliferation rate at the time point t. The curve in Fig. 22 has a sharp rise
in the beginning and this gradually slows down towards the end. With less number of 19 individuals,
the curve is different compared to that with 47 individuals shown in Fig. 10

martingale difference. We may obtain a bound on extreme fluctuation of observations

from response curve from the assumption that the errors are martingale differences.
The following moment bound of Dasgupta (1993) holds for general stochastic
processes. This includes martingales as a special case. Below we provide a slightly
modified version of the result.
14 R. Dasgupta

-0.0020
Proliferation rate of score/day
-0.0028 -0.0026 -0.0024 -0.0022

0 200 400 600 800 1000

Time (day)

d
Fig. 23 Proliferation rate of 18 persons IQ score: wt. exp(−x); spline. Proliferation rate dt log y
over 9 time points for IQ score y is computed like Fig. 10, with exponentially decaying weights
attached to empirical slopes computed from data pairs at different time points with respect to a fixed
time point t of interest in Fig. 23. Weighted mean of the empirical slopes is considered at derivative
stage of computation and the package smooth.spline with spar = 0.01 at smoothing stage in SPlus
provides the proliferation rate at the time point t. A sharp rise of proliferation rate is seen in the
beginning, which gradually slows down and then a downward trend of proliferation rate is seen.
The rate of fall is sharper towards the end
Proliferation rate of score/day
-0.0010
-0.0020
-0.0030

0 200 400 600 800 1000

Time (day)

d
Fig. 24 Proliferation rate of 19 persons Math score: wt. exp(−x); spline. Proliferation rate dt log y
over 9 time points for Math score y is computed based on a technique proposed in Dasgupta (2015),
with exponentially decaying weights attached to empirical slopes computed from data pairs at
different time points with respect to a fixed time point t of interest in Fig. 24. Weighted mean of the
empirical slopes is considered at derivative stage of computation and the package smooth.spline
with spar = 0.0001 at smoothing stage in SPlus provides the proliferation rate at the time point t. A
little bit downward trend is seen in the beginning reaching a minimum near 80 days, then the curve
gradually moves up fast to reach a high value at time t = 700 days. The curve reaches stability at
the far end. Rise of proliferation rate is slow towards the end
Longitudinal Studies on Mathematical Aptitude and Intelligence Quotient … 15

Fig. 25 Lowess on median of growth band (Math + IQ) of 47 individuals. We now examine the
data with assumption of symmetric error dropped. Instead, we consider median to calculate the
response curve. A nonparametric growth curve of total score by lowess regression on a robust
estimate of central tendency as median scores at time points fixed on dates of interview in the
band, with f = 2/3 is shown in Fig. 25. The curve dispenses off the stringent assumption that error
components are symmetric. The curve has an initial upward trend for a while, then it goes down
and reaches a minimum when initial boost from interaction with interviewer erodes; then the curve
gradually shows an upward trend over time

Fig. 26 Lowess on median of growth band (IQ) of 47 individuals. Growth curve of IQ score by
lowess regression on median of scores at time points fixed on dates of interview in the band, with
f = 2/3 is shown. Elevation of this curve is a bit higher than the previous graph on total score.
Like in the previous curve, this has initial upward trend for a while, then it goes down and reaches
a minimum when initial boost from interaction with interviewer possibly erodes; then the curve
gradually shows an upward trend over time. The pattern of the curve in Fig. 26 is not similar to that
shown in Fig. 7, indicating that the assumption of symmetric error component for IQ scores may
not hold

TheoremA Let {X i , i ≥ 1} be a stochastic process  with E[sgn(Si−1 )X i | |Si−1∗|] ≤

∗
0, E( i=1 n
±X i )2 ≤ nβ2,n , where Si = ij=1 X j , γν,n = E|X n |ν , βν,n =
max1≤ j≤n γν, j . If the l.h.s. of (1) is finite, then for ν ≥ 2

E|Sn |ν ≤ cν n ν/2 βν,n

∗
, where cν = [2(ν − 1)δ]ν/2 (1)
ν
and for large n, δ ≈ (1 + 2n
).
16 R. Dasgupta

Fig. 27 Lowess on median of growth band (Math) of 47 individuals. Growth curve of Math score
by lowess regression on median of scores at time points fixed as dates of interview in the band, with
f = 2/3 is shown in Fig. 27. Elevation of this curve is lower than the previous two graphs. Like
the previous curve, this has initial upward trend for a while. Then it goes down further compared to
Fig. 9 and reaches a minimum when initial boost from interaction with interviewer possibly erodes;
then the curve gradually shows an upward trend over time. The pattern is similar to that for Fig. 9
0.001
Proliferation rate of score/day
0.0
-0.001
-0.002
-0.003
-0.004

0 200 400 600 800 1000

Time (day)

Fig. 28 Proliferation rate of 47 persons total score (med): wt. exp(−x); spline. Proliferation rate
d
dt log y over 11 time points for total score y is computed like Fig. 10, with exponentially decaying
weights attached to empirical slopes computed from data pairs at different time points with respect
to a fixed time point t of interest in Fig. 28. Weighted mean of the empirical slopes is considered at
derivative stage of computation and the package smooth.spline with spar = 0.0001 at smoothing
stage in SPlus provides the proliferation rate at the time point t. A downward trend in the beginning
is seen and then a typical sharp rise is seen to reach a maximum at t = 700 days, the peak value
slightly crossing the initial value. A bit downward trend is seen towards the end to reach stability
of proliferation rate
Longitudinal Studies on Mathematical Aptitude and Intelligence Quotient … 17

0.002
Proliferation rate of score/day
0.001
0.0
-0.003 -0.002 -0.001

0 200 400 600 800 1000

Time (day)

Fig. 29 Proliferation rate of 47 persons IQ score (med): wt. exp(−x); spline. Proliferation rate
d
dt log y over 11 time points for IQ score y is computed based on a technique proposed in Dasgupta
(2015), with exponentially decaying weights attached to empirical slopes computed from data pairs
at different time points with respect to a fixed time point t of interest in Fig. 29. Weighted mean of
the empirical slopes is considered at derivative stage of computation and the package smooth.spline
with spar = 0.0001 at smoothing stage in SPlus provide the proliferation rate at the time point t. A
steep fall in the beginning is seen to reach a minimum at t = 250 days and then a rise upto t = 700
days is seen with peak value less than the initial value of proliferation rate seen at start. Finally
stability at the far end is seen
0.002
Proliferation rate of score/day
0.0
-0.002
-0.004

0 200 400 600 800 1000

Time (day)

Fig. 30 Proliferation rate of 47 persons Math score (med): wt. exp(−x); spline. Proliferation rate
d
dt log y over 11 time points for Math score y is computed like Fig. 10, with exponentially decaying
weights attached to empirical slopes computed from data pairs at different time points with respect
to a fixed time point t of interest in Fig. 30. Weighted mean of the empirical slopes is considered at
derivative stage of computation and the package smooth.spline with spar = 0.0001 at smoothing
stage in SPlus provides the proliferation rate at the time point t. A sharp fall in the beginning is
seen reaching a minimum at t = 250 days and then a sharp rise of the curve is seen, with rate
approximately 0.002 at t = 700 days, to cross the initial value of rate. Ability to learn mathematics
is commendable in individuals. Finally, the proliferation rate shows a downward trend to reach
stability at the far end
18 R. Dasgupta

Fig. 31 Lowess on mean of growth curves (Math + IQ) of 47 individuals. Although not robust,
mean is a natural estimate in sample for central tendency. Instead of median, we consider lowess
with f = 2/3 on mean of growth curves in Fig. 31. Rise of the curve for total score is seen at the
beginning; it then reaches a minimum and gradually rises again till the end

Fig. 32 Lowess on mean of growth curves (IQ) of 47 individuals. In this lowess curve with f = 2/3,
the fall from high initial value is steep. Then the curve in Fig. 32 rises again towards a high value
of 66.95%

Fig. 33 Lowess on mean of growth curves (Math) of 47 individuals. In Fig. 33, unlike Fig. 32 the
fall is not that steep in this lowess curve with f = 2/3 for Math score. After a fall, the curve rises
again to reach a value of 47.80%. This is lower than the corresponding value 66.95% in IQ, shown
in the previous figure
Longitudinal Studies on Mathematical Aptitude and Intelligence Quotient … 19

0.001
Proliferation rate of score/day
0.0
-0.001
-0.002
-0.003

0 200 400 600 800 1000

Time (day)

Fig. 34 Proliferation rate of 47 persons total score (mean): wt. exp(−x); spline. The proliferation
rate curve in Fig. 34 with spar = 0.0001 mimics the pattern of Fig. 28 where median is considered in
place of mean. The patterns are very similar in either case of mean or median being used to estimate
the growth curve of total score, the similarity of Figs. 27 and 31 is reflected in the similarity of
proliferation rate curves computed under different assumptions

Fig. 35 Proliferation rate of

0.001

47 persons IQ score (mean):

wt. exp(−x); spline.
Proliferation rate of score/day
0.0

Proliferation rate with spar

= 0.0001 in Fig. 35 with
-0.001

mean values for IQ rises to a

higher level than the initial
-0.002

proliferation rate at start,

unlike that of Fig. 29 with
-0.003

median of IQ scores where

initial value of the rate is the
highest
-0.004

0 200 400 600 800 1000

Time (day)

An extra term 2 in cν above appears due to the fact that expectation of max-
imum of the terms in (2.2) of Dasgupta (1993) is bounded above by sum of the
expectations in (2.3) therein, leading to an extra factor 2; i.e., the correct expression
is E max(|Sn |ν−2 X n2 , |Sn∗ |ν−2 X n2 ) < E(|Sn |ν−2 X n2 + |Sn∗ |ν−2 X n2 ). The bound is useful
in estimating remainder from the main part in nonlinear statistics that arise in many
situations related to limit theorems, see, e.g., Dasgupta (1994). The modification
does not affect the results of Dasgupta (1994) as L is a generic positive constant
therein. 
Consider Mi = ij=1 y j , where M is a martingale and y s are the martingale
differences. For a martingale M with finite νth moment, one may write From Doob’s
inequality
20 R. Dasgupta

0.0010
Proliferation rate of score/day
0.0005
0.0
-0.0005
-0.0010

0 200 400 600 800 1000

Time (day)

Fig. 36 Proliferation rate of 47 persons math score (mean): wt. exp(−x); spline. Proliferation rate
with spar = 0.0001 in Fig. 36 with mean values for Math scores has a sharp downward trend in the
beginning. Then the curve rises again to achieve a peak with lower value than the initial value at
start, before stability is achieved; this pattern is unlike Fig. 30 with median score, where the peak is
higher than the initial value

P( max |Mi | ≥ c) ≤ c−ν E|Mn |ν (2)

1≤i≤n

Thus, if a finite ν(≥ 2) th order moment of martingale differences y j exist, i.e.,

sup E|y j |ν < ∞

j≥1

then from (1) and (2) one may write

P( max |Mi | ≥ λn 1/2 ) ≤ λ−ν cν βν,n

∗
(3)
1≤i≤n

Next consider the case where all the moments of martingale differences exist, but
the moment generating function need not necessarily exist. Consider the following
moment bounds for the martingale differences y j .
Type 1:
α
sup E|y j |ν ≤ Lew0 ν (4)
j≥1

∀ν > 1, and for some L > 1, where w0 > 0, α > 1. In what follows L(> 1) is a
generic constant. The above condition is equivalent to
Another random document with
no related content on Scribd:
 Music for the New Church

As the kingdom changed its king at the death of each monarch, the
country swayed from Catholicism to Protestantism and back again,
and many a poet and musician lost his head or was burnt at the stake
because he wrote for the Protestant Church. In the case of Marbeck
who had made music for the Book of Common Prayer, he just
escaped death for the crime of writing a Bible concordance (an
index)!
Before Wynken de Worde’s song book came out, William Caxton,
the great printer, published a book called Polychronicon by Higden.
In this, was an account of Pythagoras and his discovery of tone
relations (Chapter IV); this proves the great interest in England for
the science, as well as the art, of music.
In Frederick J. Crowest’s book, The Story of the Art of Music, he
tells very simply the state of music in England at this time:
“When the adventurous Henry VIII plunged into and
consummated (completed) the reformation scheme, it was at the
expense of considerable inconvenience to musicians obliged,
perforce, to change their musical manners as well as their faith. In
double quick time the old ecclesiastical (church) music had to be cast
aside, and new church music substituted.... This meant pangs and
hardships to the musicians, possibly not too industrious, accustomed
to the old state of things. Simplicity, too, was the order, a change that
must have made musicians shudder when they, like others before
them, from the time of Okeghem, had regarded the Mass as the
natural and orthodox (correct) vehicle for the display of the
contrapuntal miracles they wrought.”
Now the Mass became the “Service,” and the motet was turned
into the “Anthem,” which we still use in our churches. Most of the
famous composers of the 16th and 17th centuries in England wrote
for the new Anglican or Protestant Church, and made the new music
lovely indeed. Many of them were organists or singers in the Chapel
Royal, so they had been well prepared for their work.
To make this new music different from the old, the writers were
ordered to fit every syllable with a chord (in the harmonic style). In
the old counterpoint, of course, the words were somewhat blurred.
These experiments with chords did much to free music for all time.
One of the earliest of the church composers is Thomas Tallis
(about 1520–1585), a “Gentleman of the Chapel Royal” and father of
English cathedral music. Through his long career, Tallis followed the
different religions of the rulers from Henry VIII to Elizabeth, writing
Catholic music or Protestant as was needed. You see he liked his
head, so he changed his music with each new monarch. He, like some
of the composers of the Netherlands school, wrote a motet for forty
voices.
He shared with his pupil, William Byrd, the post of organist of the
Chapel Royal, and together they opened a shop “to print and sell
music, also to rule, print and sell music paper for twenty-one years”
under a patent granted by Queen Elizabeth to them only. How
successful the two composers were in the publishing business is not
stated, but at least they could publish as many of their own works as
they cared to! After Tallis’ death, in 1585, for a while Byrd ran the
shop alone, and published a collection of Psalms, Sonets, and Songs
of Sadness and Pietie. In this was written “Eight reasons briefly set
down by the Author (Byrd) to persuade every one to learn to sing” to
which he added:
Since singing is so goode a thing
I wish all men would learne to sing.
 Famous Old Music Collections

England was the land of famous music collections in the 16th and
17th centuries. The first of these by Byrd was a book of Italian
Madrigals with English words, Musica Transalpina, (Music from
across the Alps). The entire title was (Don’t laugh!): “Musica
Transalpina; Madrigals translated of foure, five, and sixe parts,
chosen out of diuers excellent Authors, with the first and second part
of La Virginella, made by Maister Byrd vpon two Stanz’s of Ariosto
and brought to speak English with the rest. Published by N. Yonge, in
fauer of such as take pleasure in Musick of voices. Imprinted at
London by Thomas Easy, the assigne of William Byrd, 1588. Cum
Privilegio Regise Maiestatis (With permission of her Royal
Majesty).” A long title and one that would not make a book a
“bestseller” today! Do notice how they mixed u’s and v’s and put in
e’s where you least expect them!
There were fifty-seven madrigals in the long titled collection
including the two by “Maister Byrd”; the others were by the Italian
and Netherland madrigal writers, such as Palestrina, Orlandus
Lassus and Ferabosco, a composer of masques and madrigals, who
lived for years in England.
Byrd’s compositions in this work mark the beginning of the great
English school of madrigals, which were so lovely that this period
(1560 to 1650) was called the “Golden Age.”
 The Golden Age of Madrigals

Now the madrigal becomes the great English contribution to

music. It was a part-song in free contrapuntal style and the music
was made to fit the words. For the first time, secular music was held
in great honor, and prepared the way for arias, dramatic solos and
original melodies.
After Byrd and Edwards, came other madrigal writers: Thomas
Morley, John Dowland, George Kirby, Thomas Ford, Thomas
Ravenscroft, Orlando Gibbons and others.
While the madrigal was being written in England and elsewhere,
the part-song was being written in Germany. It was the companion
of the chorale, as the madrigal was the secular partner of the motet.
The chorale was written for part singing, had a continuous melody
and the same air was used for all stanzas. In this the church modes
were never used, yet, it is baffling sometimes to tell a madrigal from
a part-song.
In Italy the villanella, or villota is a part-song. In France it was the
chanson, in England it was the madrigal or the glee.
 “The Triumphs of Oriana”

Monarchs, besides ruling the country, inspired poets and

composers from earliest times, and Queen Elizabeth was no
exception. The Triumphs of Oriana is a collection of madrigals by
many English composers in praise of Queen Elizabeth, made by
Thomas Morley. Because William Byrd does not appear in it, it looks
as if this collection had been published to show Byrd that the English
could write good madrigals, too. Anyhow, it definitely proves that the
English madrigals are as charming as the French, Italian or Flemish.
There is a copy of the original edition in the British Museum.
 Maister Byrd Gives Advice

In 1611, an important work of Byrd’s appeared called Psalms,

Songs and Sonets: some solemne, others joyfull, framed to the life of
the words: Fit for Voyces or Viols. In the dedication, the composer
gives this good advice: “Onely this I desire; that you will be but as
carefull to heare them well expressed, as I have beene both in the
composing and correcting of them. Otherwise the best Song that euer
was made will seeme harsh and vnpleasant.... Besides a song that is
well and artificially made cannot be well perceived nor vnderstood at
the first hearing, but the oftner you shall heare it, the better cause of
liking it you will discouer; and commonly the Song is best esteemed
with which our eares are best acquainted.”
Over the door of the music hall in Oxford University, is a canon (or
round) for three voices, said to have been written by William Byrd.
Some day, if you have not already seen it, you will have the thrilling
experience of visiting the venerable college, and you may remember
to look for this canon.
 Ladies of the Realm Play Virginals

As today we consider no home complete without a piano (or

pianoforte which is its real name), so in the 16th and 17th centuries
we would have found a little key board instrument so small that it
could easily be swallowed whole by one of our grand pianos, and you
would never know where it had disappeared! It was known by several
names,—spinet, clavecin, and virginal or virginals. Another
instrument belonging to the same family and period is the
harpsichord, which is more like our grand piano in shape. But later
we will tell you more of the pianoforte’s family tree, and of its tiny
but important grand-parent.
It was quite the proper thing for all the ladies of the realm to play
the virginals, and the Queens, Mary, Elizabeth, and Mary, Queen of
Scots, were excellent performers.
The very first music printed for the virginals in England was called
Parthenia (from the Greek word Parthenos, meaning unmarried
woman or virgin). The printed title also tells us that it was
“composed by three famous masters, William Byrd, Dr. John Bull
and Orlando Gibbons Gentilmen of his Majesties most illustrious
Chappell.” There are twenty-one pieces from the old dances which
formed the Suites, of which you will soon hear,—Preludiums,
Pavanes, Galiardes, a Fantasia, and one The Queene’s Command. It
was published in 1611, on staves of six lines, instead of five, as we
use, and it was the first musical work engraved on copper plates!
 More Famous Collections

Another most valuable collection was for many years called Queen
Elizabeth’s Virginal Book, but is now the Fitzwilliam Virginal Book,
and the original manuscript is in the Fitzwilliam museum at
Cambridge. It was supposed to have been “Good Queen Bess’” book,
but it was not, as some of its compositions were composed after her
death. It is not known who copied 220 pages of music, but it may
have been a wealthy Roman Catholic, Francis Tregian, who spent
twenty-four years in prison on account of his religious faith. This
name, abbreviated or in initials, is found in several places in the
manuscript. An edition in our notation has been made by J. A. Fuller
Maitland and W. Barclay Squire. Many of the old songs of English
minstrelsy are found among the numbers, and they were arranged
for the instrument by the famous composers of that day. There are
also original compositions as well as “ayres” and variations. Among
the composers we find Dr. John Bull, Thomas Morley, William Byrd,
Orlando Gibbons, Giles Farnaby, Richard Farnaby, Thomas Tallis,
Jan Sweelinck, the Dutch organist, and many others. Here are some
of the quaint titles: St. Thomas’ Wake, King’s Hunt, The Carman’s
Whistle, The Hunt’s Up, Sellonger’s Round, Fortune My Foe, Pawles
Warfe, Go from My Window, Bonny Sweet Robin, besides many
Pavanes, Galiardes, Fantasias, and Preludiums.
There is also a collection of Byrd’s virginal music called My Lady
Nevell’s Booke. Lady Nevell may have been a pupil of Byrd. There are
two collections of this same kind at Buckingham Palace, the home of
the King of England,—Will Forster’s Virginal Book and Benjamin
Cosyn’s Virginal Book. In the index of the latter, we read: “A table of
these Lessons following made and sett forth by Ben Cos.” In all, he
copied more than 90 compositions!
Later came John Playford, music publisher, whose first musical
publication, The English Dancing-Master (1650), contains almost a
hundred old folk tunes. Select Musical Ayres appeared three years
later, and is a typical 17th century song collection of first-class poems
by Jonson, Dryden and others set to music by well-known
composers. His book on the theory of music, used for almost a
century, contained “lessons” for the viol, the cithern and flageolet.
His Dancing Master, a collection of airs for violin for country
dances, has brought to us many popular ballad tunes and dance airs
of the period.
In these collections we often find the names Fancies, Fantazia, or
Fantasies, a type of composition that grew out of the madrigal and
led to the sonata. It was the name given to the first compositions for
instruments alone like the ricercari of the Italians, which were
original compositions and not written on a given subject (called in
England “ground”), or on a folk song. The Fancies were sometimes
written for the virginal, and sometimes for groups of instruments
such as a “chest of viols” or even five cornets(!).
 The Chest of Viols

“Chest of Viols” may sound queer to you, but it isn’t! It was the
custom in England at that time for people to have collections of
instruments in or out of chests. So, when callers came they could
play the viol, instead, probably of bridge! You can read about these
interesting old days in Samuel Pepys’ Diary. He played the lute, the
viol, the theorbo, the flageolet, the recorder (a kind of flute) and the
virginal, and he was the proud owner of a chest of viols. He always
carried his little flageolet with him in his pocket, and he says that
while he was waiting in a tavern for a dish of poached eggs, he played
his flageolet, also that he remained in the garden late playing the
flageolet in the moonlight. (Poetic Pepys!)
Thomas Morley, Byrd’s pupil, who was made a partner in the
publishing house after Tallis’s death, wrote his madrigals for virginal,
and a collection called First Book of Consort Lessons for Six
Instruments, Lute, Pandora, Cittern (an old English form of guitar),
Bass Viol, Flute, and Treble Viol, and much sacred music. He also
wrote a Plaine and Easie Introduction to Practical Musick, a book of
great value and interest to musicians for the last three centuries, for
it is a mirror of his time and of his fellow composers.
He tells of a gentleman, who, after dinner, was asked by his
hostess to sing from the music she gave him. It was the custom in
England to bring out the music books after dinner and for the guests
to play and sing, as we wind up our graphophones and switch on the
radio. The gentleman stammeringly declared that he could not sing
at sight and “everyone began to wonder; yea, some whispered to
others, demanding how he was brought up.” He was so ashamed of
his ignorance that he immediately took music lessons to remedy his
woeful lack of culture. This proves that musical education was not
looked upon as a luxury but a necessity in the 17th century.
Truly, it was a musical era, this time of Morley and Byrd! Fancy
playing, while waiting for the barber, the viol, flute, lute, cittern, or
virginal left for that purpose. Yet what would our dentist do today if
he had to listen to a saxophone and jazz chorus from his waiting
room? In those days, too, there was always a bass viol left in a
drawing-room for the guest, to pass the time, waiting for the host to
appear. Think of all the practising you could do waiting for the busy
dentist or eternally late hostess!
The children of people who were poor, were taught music to make
them fit to be “servants, apprentices or husbandmen.” Laneham, a
groom who had been brought up in the royal stable, was advanced to
the post of guarding the door of the council chamber and this is how
he described his qualifications for the job: “Sometimes I foot it with
dancing; now with my gittern, and else my cittern, then at the
virginals (ye know nothing comes amiss to me); then carol I up a
song withal; that by-and-by they come flocking about me like bees to
honey; and ever they cry, ‘Another, good Laneham, another!’” (From
The Story of Minstrelsy by Edmundstoune Duncan.)
 Shakespeare and Music

This was the day in which Shakespeare lived, and from his plays
we get a very good idea of the popular music of his time, for he used
bits of folk songs and old ballads. It was a Lover and his Lass from
As You Like It was set to music by Thomas Morley, and is one of the
few songs written to Shakespeare’s words in his own day that has
come down to us. In Twelfth Night there is O Mistress Mine, Hold
thy Peace, Peg-a-Ramsey, O, London is a Fine Town, Three Merry
Men be We, and the Clown’s song:
Hey! Robin, jolly Robin,
Tell me how thy lady does, etc.

In the Winter’s Tale, As You Like It, The Tempest, Merchant of

Venice, Hamlet, Othello are folk songs that are very well known and
loved. Two songs from The Tempest, Where the Bee Sucks and Full
Fathoms Five, were set to music by a composer, Robert Johnson,
who lived at the same time as Shakespeare, but was not as famous as
Morley, who also lived then. O, Willow, Willow, sung by Desdemona
in Othello is one of the most beautiful and saddest folk songs we
know.
One Shakespeare song has been made famous by the beautiful
music which the great German song writer, Schubert, wrote to it. It is
from Two Gentlemen of Verona and is called Who is Sylvia?
Many of the English composers of the 17th and 18th centuries such
as Henry Purcell and Dr. Arne made music for the Shakespeare
songs because they were so lovely and so well written that they
almost sang themselves; this we call lyric verse.
Thomas Weelkes (1575?–1623) whose madrigals were included in
The Triumphs of Oriana, also wrote many Fancies for Strings which
were the ancestors of the string quartets, the highest type of music.
 Cryes of London

Several composers of this period, Thomas Weelkes, Orlando

Gibbons (1583–1625) and Richard Deering (1580?–1630) wrote
pieces using the old “Cryes of London” as their themes. Each trade
had its own song, and the street pedlars used these tunes just as the
fruit vendors, old-clothes men, and flower vendors cry their wares in
our streets today. There is this difference, however; the street cries of
today are mere noise, while the old “Cryes of London” were
interesting and usually beautiful songs. Cherry Ripe is one of them,
and Campion used it in 1617 in his famous old song, There is a
Garden in Her Face. Some of the composers made rounds and
catches based on the “Cryes,” and Weelkes in his Humorous Fancy
used the songs of the chimney-sweep, the bellows-mender, and the
vendors of fruit, fish and vegetables. In telling about this “fancy,”
Frederick Bridge, a British composer and professor of music in
Gresham College, says: “The Fancy at one point leaves its regular
course, and for a few bars a delightful dance tune is introduced, to
the words, whatever they mean, ‘Twincledowne Tavye.’ It is as if the
vendors of fish, fruit and vegetables met in the street and had a bit of
a frolic together.” Bridge also says that he thinks all lovers of
Shakespeare will be glad to make the acquaintance of the music of
the “Cryes of London” which saluted the poet’s ears in his daily walk.
Orlando Gibbons called his composition on the “cryes,” a
Burlesque Madrigal, and beside the cries, he has used in one of the
inner parts for viol, an old plain-song melody, a form used very often
by the Italian madrigalists of the 16th century. Richard Deering’s
Humorous Fancy, The Cryes of London, is the most elaborate of the
three we have mentioned, having among many other tradesmen’s
songs, those of the rat-catcher (this makes us think of Browning’s
Pied Piper of Hamelin), the tooth-drawer, and the vendor of garlic.
 Some Famous Composers

Orlando Gibbons was one of the composers of Parthenia. But he is

famous as a composer of sacred music, in fact, he is looked upon as
the greatest composer of the English contrapuntal school. His
anthems are still sung in the English Cathedrals, and one of them
made for James I, was sung, in part, at the coronations of both
Edward VII and George V, and is now called the Abbey Amen.
Gibbons, Byrd and Bull were very fine organists. Gibbons was
organist of Westminster Abbey, and we are told by a writer of his
own day that “the organ was touched by the best finger of that age,
Mr. Orlando Gibbons.”
Dr. John Bull (1563–1628) was brought up, as were many of the
young English musicians, as one of the “Children of the Chapel Royal
Choir.” Later he became organist and player to King James I. Bull
left England, entered the service of a Belgian archduke, was organist
at the Antwerp Cathedral, and when he died in 1628, he was buried
there. In the University of Oxford, where Bull took his degree as
Doctor of Music, is his portrait around which is written:
The Bull by force in field doth rayne
But Bull by skill good-will doth gaine.

John Milton, father of the great poet, was an important composer

of this period. It is well known that his famous son was very fond of
music, was a good musician himself, and had many friends among
these composers and musicians.
The music for Milton’s famous Masque, Comus, was written by
Henry Lawes (1595–1662) and was first produced in 1635. Lawes
studied with an English composer named John Cooper who lived for
so many years in Italy, that his name was translated into Giovanni
Coperario. He turned the thoughts of his pupil to composing music
for the stage, instead of church music. It looks as if Milton had been
a pupil of Lawes, and had written Comus specially for him.
Lawes played a very amusing joke upon the concert-goers. At that
time, as now, many thought that the music of other countries, and
songs in foreign languages were better than their own. While Lawes
himself knew the Italian music very well, he was eager to compose
music that should be truly English. In the preface to his Book of
Ayres he confessed: “This present generation is so sated with what’s
native, that nothing takes their ears but what’s sung in a language
which (commonly) they understand as little as they do the music.
And to make them a little sensible of this ridiculous humor, I took a
Table or Index of old Italian Songs and this Index (which read
together made a strange medley of nonsense) I set to a varyed Ayre,
and gave out that it came from Italy, whereby it has passed for a
rare Italian song.” (Quoted from Bridge’s Twelve Good Musicians.)
Lawes helped to compose a work that is looked upon as the first
English opera, The Siege of Rhodes. This was played during the time
of Oliver Cromwell and the Commonwealth, and in this piece for the
first time in England, women appeared upon the stage.
A year after the Commonwealth was overthrown, Henry Lawes
died and was buried in Westminster Abbey, but the spot where his
body lies is not known.
From 1641 to 1660, music must have had a hard time for this was
the period of the Commonwealth, when the country was going
through all the horrors of civil war, and Cromwell’s soldiers
destroyed many things of great artistic value, that could never be
replaced. Among them were the works of art found in the wonderful
old English cathedrals, including organs and musical manuscripts.
At Westminster Abbey, the Roundheads (the name given to Oliver
Cromwell’s soldiers) “brake down the organs for pots of ale.”
Matthew Locke (1630?–1677) is looked upon as the “Father of
English Opera.” He wrote the music for Psyche and The Tempest
(1673). Another Shakespeare play to which Locke wrote the music
was Macbeth.
 Captain Cooke’s Choir Boys

Immediately after the Restoration, the Chapel Royal Choir was

reorganized. For centuries it had been the great school of music for
the sons of both rich and poor, and had produced nearly all the
English musicians. Captain Henry Cooke, the first chapel master of
the new choir, seems to have picked out unusually gifted children,
some of whom wrote anthems while they were still in the Choir, and
afterwards became very famous composers, among them John Blow,
Pelham Humphrey and the great Henry Purcell. The Captain
evidently knew how to train his boys!
Pelham Humphrey, having attracted the attention of the King, was
sent to Paris to study with the famous opera composer, Lully. The
effect of this study was felt in English music, as Humphrey was
Purcell’s master at the Chapel Royal, after the death of the good
Captain Cooke, and he introduced his new ideas to his talented little
choir boys and musical friends. Samuel Pepys says that the visit to
Paris made a snob of “little” Pelham Humphrey: “He is an absolute
Monsieur, full of form and confidence and vanity, and disparages
everything and everybody’s skill but his own. But to hear how he
laughs at all the King’s Musick here, ... that they cannot keep time
nor tune nor understand anything.”
Dr. John Blow (1648–1708) composed Anthems while still a choir
boy, and at twenty-one was organist of Westminster Abbey. In 1674
he was Master of the Children of the Chapel Royal, and became its
organist as well, without giving up his post at Westminster. During
part of the time Purcell was at Westminster, and Blow was Almoner
and Master of the choristers in St. Paul’s Cathedral. Think of filling
three of the greatest positions in musical London at the same time!
He wrote an Anthem, I was Glad, for the opening of St. Paul’s
Cathedral in 1697.
He wrote many church compositions, masques, and pieces for
harpsichord.
Purcell called him “one of the greatest masters in the world.” Like
Monteverde, he tried out new effects in harmony and made new
combinations which have since been called “crude,” but were signs of
a musical daring and understanding that belong only to very gifted
musicians.
He died in 1708 and is buried in Westminster Abbey.
 Maister Purcell

The last of the great 17th century English composers, and the
greatest of them all, is Henry Purcell (1658–1695). His father was a
well-known musician, and the uncle, who brought him up, was also a
musician, so the young boy heard much music in his own home, and
no doubt knew many composers.
Sir Frederick Bridge in Twelve Good Musicians tells us that the
Purcell family came from Tipperary in Ireland and that Henry’s
father and uncle were Gentlemen in the Chapel Royal in London.
Henry began his music studies at the age of six, for he, too, was one
of “Captain Cooke’s boys,” and when he was twelve years old,
“Maister Purcell” wrote a composition in honor of “His Majestie’s
Birthday.”
The young Purcell, sometimes called the “English Mozart,” gained
much from Pelham Humphrey who told him of Lully in France. After
Humphrey’s early death (he was only twenty-seven), Purcell studied
with Dr. Blow, and the two musicians were devoted comrades. Their
tombs lie close together near the old entrance of the organ loft,
where they must have spent many hours of their lives.
Matthew Locke was also a friend of Purcell’s, and probably did
much to interest the young composer in the drama, for in spite of his
early church training, Purcell’s greatest offering to English music was
his opera writing. While Purcell’s are not operas in our sense of the
word, they are the nearest thing to them that England had, before the
Italians came with theirs in the 18th century. He wrote music to
masques and plays, several of which were even called operas, yet
only one really was an opera. Purcell’s music “was so far in advance
of anything of the sort known in any part of Europe in his day, in
point of dramatic and musical freedom and scenic quality, that one
can only regret his early death’s preventing his taking to opera
writing on a larger scale.” (W. F. Apthorp.) Among the things he put
to music were the plays of Dryden and of Beaumont and Fletcher.
Purcell was one of the first English composers to use Italian
musical terms, like adagio, presto, largo, etc. He was also one of the
first composers to write compositions of three or four movements for
two violins, ’cello and basso continuo, a part written for harpsichord
or sometimes organ as an accompaniment to the other instruments.
The name of this style of composition also came from the Italian, and
was called Sonata. The first sonatas were composed by Italians. The
word Sonata comes from an Italian word suonare which means to
sound, and was first given to works for instruments. Another form of
composition is the Cantata, from cantare which means to sing. It is a
vocal composition with accompaniment of instruments, a direct
descendant of the motet and madrigal, and of the early oratorios.
The Toccata, too, comes from the Italian toccare, meaning to
touch, and was originally a work for instruments with keyboards.
The Italian language gave us our musical names and terms, because
Italian music was the model of what good music should be, and
England, France and Germany copied Italian ways of composing.
Everyone uses the Italian terms for musical expressions so that all
nationalities can understand them.
When Purcell was only 17 years old, he composed an opera to be
played by young ladies in a boarding school. This was Dido and
Æneas, and it is so good that few writers on musical subjects believe
that it was written in his youth.
In every branch of composition in which Purcell wrote, he excelled.
His church music is the finest of his day, his chamber music and his
operas are looked upon as works of genius. In fact, he is still
considered the most gifted of all English composers.
He was only 37 when he died, and was a very great loss to the
growth of English music.
Music Comes of Age