Vox Latina A Guide to the Pronunciation of

Classical Latin 2nd ed. Edition William Sidney

Allen pdf download
[Link]
edition-william-sidney-allen/

★★★★★ 4.6/5.0 (36 reviews) ✓ 204 downloads ■ TOP RATED

"Great resource, downloaded instantly. Thank you!" - Lisa K.

DOWNLOAD EBOOK
 Vox Latina A Guide to the Pronunciation of Classical Latin
2nd ed. Edition William Sidney Allen pdf download

TEXTBOOK EBOOK EBOOK GATE

Available Formats

■ PDF eBook Study Guide TextBook

EXCLUSIVE 2025 EDUCATIONAL COLLECTION - LIMITED TIME

INSTANT DOWNLOAD VIEW LIBRARY

Instant digital products (PDF, ePub, MOBI) available
Download now and explore formats that suit you...

Telling to Live Latina Feminist Testimonios Latin America

Otherwise Languages Empires Nations Latina Feminist Group

[Link]
testimonios-latin-america-otherwise-languages-empires-nations-latina-
feminist-group/
[Link]

Gran Cocina Latina The Food of Latin America 1st Edition

Maricel E. Presilla

[Link]
america-1st-edition-maricel-e-presilla/

[Link]

A Field Guide to the Birds of Peninsular Malaysia and

Singapore 2nd Edition Allen Jeyarajasingam

[Link]
peninsular-malaysia-and-singapore-2nd-edition-allen-jeyarajasingam/

[Link]

The Field Guide to Understanding Human Error 3rd Edition

Sidney Dekker

[Link]
error-3rd-edition-sidney-dekker/

[Link]
The Pronunciation of English A Course Book 2nd Edition
Charles W. Kreidler(Auth.)

[Link]
book-2nd-edition-charles-w-kreidlerauth/

[Link]

A companion to Woody Allen Allen

[Link]

[Link]

The Britannica guide to geometry 1st ed Edition William L.

Hosch

[Link]
edition-william-l-hosch/

[Link]

Handbook of classical mythology William Hansen

[Link]
hansen/

[Link]

Latin Literature Classical Foundations Susanna Morton

Braund

[Link]
susanna-morton-braund/

[Link]
VOX LATINA
VOX LATINA
A GUIDE TO
THE PRONUNCIATION OF
CLASSICAL LATIN

BY

W. SIDNEY ALLEN
Professor Emeritus of Comparative Philology in the
University of Cambridge

Second edition

j CAMBRIDGE
UNIVERSITY PRESS
PUBLISHED BY THE PRESS SYNDICATE OF THE UNIVERSITY OF CAMBRIDGE
The Pitt Building, Trumpington Street, Cambridge, United Kingdom
CAMBRIDGE UNIVERSITY PRESS
The Edinburgh Building, Cambridge CB2 2RU, UK
40 West 20th Street, New York, NY 10011-4211, USA
10 Stamford Road, Oakleigh, VIC 3166, Australia
Ruiz de Alarcon 13, 28014 Madrid, Spain
Dock House, The Waterfront, Cape Town 8001, South Africa

http: / /www. [Link]

© Cambridge University Press 1965, 1978

This book is in copyright. Subject to statutory exception

and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without
the written permission of Cambridge University Press.

First published 1965

Reprinted 1970
Second edition 1978
Reprinted 1989
First paperback edition 1989
Reprinted 1990, 1992, 1996, 1999, 2001

ISBN 0 521 22049 1 hardback

ISBN 0 521 37936 9 paperback
(First edition ISBN 0 521 04021 3)

Transferred to digital printing 2004

 FOREWORD TO THE
SECOND EDITION
A reprint of this book with minor corrections was issued in
1970. Now that a further printing is called for, I have taken the
opportunity to incorporate some major revisions, in order to
take account of further studies which have appeared or come to
my notice since 1965, as well as any changes or developments in
my own ideas. I have also added a number of references for the
reader interested in following up some of the more crucial or
controversial points. In the meantime there have also appeared
my Vox Graeca (C.U.P., 1968; second edition 1974; third
edition 1987) and Accent and Rhythm (C.U.P., 1973), to which
there are several cross-references (abbreviated as VG and AR
respectively).
In order to save expense and at the same time to avoid
changes in pagination from the first edition, the new material
has (as in the second edition of VG) been added as a supplement
rather than worked into the main text (which contains only
minor revisions). An obelus in the margin indicates the
existence of a relevant supplementary note.
In addition I have now included a Select Bibliography and
have appended a short account of the Latin (and English)
names of the letters of the alphabet.

Cambridge W.S.A.
July 1QJ7 (revised August 1988)
 FOREWORD TO THE
FIRST EDITION

In discussions on the subject of Latin pronunciation two ques-

tions are commonly encountered; they tend to be of a rhetorical
nature, and are not entirely confined to non-classical disputants.
First, why should we concern ourselves with the pronunciation
of a dead language? And second, how in any case can we know
how the language was originally pronounced?
In answer to the first question, it may reasonably be held
that it is desirable to seek an appreciation of Latin literature,
and that such literature was based on a living language. More-
over, much of early literature, and poetry in particular, waj
orally composed and was intended to be spoken and heard rather
than written and seen. If, therefore, we are to try and appre-
ciate an author's full intentions, including the phonetic texture
of his work, we must put ourselves as nearly as possible in the
position of the native speaker and hearer of his day. Otherwise,
however full our grammatical and lexical understanding of the
work, we shall still be missing an important element of the
contemporary experience. It is true that we can have a lively
appreciation of, say, Shakespeare, whilst reading or hearing his
work in a modern pronunciation—but in this case the two
languages are not far removed from one another, and whilst
individual sounds may have changed to some extent, the rela-
tions between them have been largely preserved; the situation
is already very different, even within English, if we go back only
as far as Chaucer. It is said that Burke used to read French
poetry as if it were English; when one considers the vowel
harmonies of a line like Hugo's ' Un frais parfum sortait des
touffes d'asphodele', one can only conclude that his apprecia-
tion must have been minimal!
We are here concerned primarily to reconstruct the educated
vii
 FOREWORD TO THE FIRST EDITION

pronunciation of Rome in the Golden Age. But it will be

necessary to take note of certain variations even within this
period, and of interest in some cases to refer to features of more
colloquial speech, and of preceding or following periods.
The degree of accuracy with which we can reconstruct the
ancient pronunciation varies from sound to sound, but for the
most part can be determined within quite narrow limits. In
some favourable cases it is possible to reconstruct such niceties
of pronunciation as it would be unreasonable to demand in
normal reading; and the present book is not so unpractical as
to suggest that more than a reasonable approximation should
then be made. But the knowledge should nevertheless be
available to the reader, so that, whatever pronunciation he in
fact adopts, he may know to what degree and in what respects
if differs from the probable original. For many of us, already
well set in our ways, it will inevitably continue to be a case of
'uideo meliora proboque; deteriora sequor'; but scholarship
surely requires that we should at least know what is known or
at any rate probable.
It is claims such as those of the preceding paragraph that
commonly evoke the second question 'How do we know?' And
there is no one simple answer to it. The kinds of evidence and
argument are various, and will become familiar in the course
of the pages that follow; but the principal types of data invoked
in phonetic reconstruction may be summarized as follows:
(i) specific statements of Latin grammarians and other authors
regarding the pronunciation of the language; (2) puns, plays
on words, ancient etymologies, and imitations of natural sounds;
(3) the representation of Latin words in other languages;
(4) developments in the Romance languages; (5) the spelling
conventions of Latin, and particularly scribal or epigraphic
variations; and (6) the internal structure of the Latin language
itself, including its metrical patterns. Our arguments will
seldom rely on one type of evidence alone, and the combina-
tions of evidence will vary from case to case. The grammarians
viii
 FOREWORD TO THE FIRST EDITION

are mostly of very late date, but their evidence is important as

confirming the continuation of features established for earlier
periods by other means; frequently also they quote the views
or practice of earlier writers; and it is a characteristic of their
profession to preserve earlier traditions long after they have
vanished from normal speech.
In view of the prevalence of the second question, it is at least
as important that the reader should be equipped with reasons
as with results; and particular attention has been paid to setting
out 'how we know what we know' in language that is, so far
as possible, free from technical complications. In the process
of reconstruction we are of course dependent on a variety of
linguistic theories and techniques, but since the present book is
not directed primarily to the linguistic specialist, no technical
terms have been used without due explanation.1 Referenpes to
the specialist literature have also been kept to a minimum; this
must not, however, be taken to minimize the debt that is owed
to a large member of books and articles, on every aspect of the
subject, over a period of roughly a century; and in particular
to such eminent overall studies as Seelmann's Die Aussprache des
Latein nach physiologisch-historischen Grundsdtzen (1885), Sommer's
Handbuch der lateinischen Laut- und Formenlehre (1914), and Sturte-
vant's The Pronunciation of Greek and Latin (1940). Two parti-
cularly useful recent works may also be specially mentioned:
Maria Bonioli's La pronuncia del latino nelle scuole daWantichitd al
rinascimento, Parte 1 (Torino, 1962), and Alfonso Traina's
Lalfabeto e la pronunzia del latino (2nd edn., Bologna, 1963).
My thanks are due to several colleagues and students for
encouragement and suggestions in the preparation of this work;
in particular to Mr A. G. Hunt, of the Department of Educa-
tion, University of Cambridge; and to Mr W. B. Thompson, of
the Department of Education, University of Leeds, who ' tried
1
The more common phonetic terms are introduced and explained in a pre-
liminary chapter, and an asterisk against the first occurrence of a term in the text
indicates that it is there discussed.
ix
 FOREWORD TO THE FIRST EDITION
out' an early draft on a number of classical school-teachers and
gave me the benefit of their comments and criticisms. I am
also grateful for the interest expressed by the Joint Association
of Classical Teachers, and by the Education Subcommittee of
the Council of the Classical Association. Lastly, I owe a special
debt to Mr R. G. G. Coleman, of Emmanuel College, Cam-
bridge, who read the whole of the final draft and made a
number of valuable comments and suggestions.
W. S. A.
CAMBRIDGE
March 1964
 CONTENTS

Foreword to the second edition page v

Foreword to the first edition vii
Phonetic Introduction i

1 CONSONANTS II

(i) Voiceless plosives, 12: t, 13; c, 14; qu, 16.

(ii) Voiced plosives, 20: g, 22; gn, 23; gu, 25.
(iii) Aspirates, 26. (iv) Nasals, 27: n, 27; m, 30.
(v) Liquids, 32: r, 32; 1,33. (vi) Fricatives, 34: f, 34;
s
> 35- ( vii ) Semivowels, 37: i, 37; u, 40. (viii) h, 43.
(ix) x and z, 45

2 VOWELS 47
(i) Simple vowels, 47: special qualities, 50; y, 52;
Old Latin e, 53; the 'intermediate vowel', 56.
(ii) Diphthongs, 60: ae and au, 60; oe, 01, 62;
ui, 62; eu, 63; ei, 63; ou, 63

3 V O W E L LENGTH 64

(i) General, 64. (ii) 'hidden quantity', 65: before

ns, nf, 65; before net, nx, 66; before x, ps, 67;
before sc, 68; 'Lachmann's Law', 68; before gn,
71; before r + consonant, 73; before final m, 74;
miscellaneous, 74. (iii) hie and hoc, 75

4 V O W E L JUNCTION 78

5 ACCENT 83

6 QUANTITY 89

Accent and quantity in classical Latin verse, 92

xi
 CONTENTS

APPENDICES

A i Selected quotations from the Latin gram-

marians and other writers page 95
2 Chronology of sources 100

B The pronunciation of Latin in England 102

C The names of the letters of the Latin alphabet 111

SUPPLEMENTARY NOTES 116

Select Bibliography 131

SUMMARY OF RECOMMENDED PRONUNCIATIONS 132
Video rem operosiorem esse quam putaram,
emendate pronuntiare.
( L E O , in D. Erasmi De recta Latini Graecique
sermonis pronuntiatione Dialogo)
 The organs of speech

B Back of tongue N Nasal cavity

E Epiglottis (drawn over P Pharynx
windpipe when swallowing) S Soft palate (velum),
F Food-passage in lowered position
G Gums (alveoli) T Tongue-tip
H Hard palate U Uvula
L Larynx, with 'Adam's apple' V Vocal cords (glottis)
M Middle of tongue W Windpipe

[After Ida C. Ward, The Phonetics of English]

XIV
 PHONETIC INTRODUCTION

(i) Syllable, vowel and consonant

In any continuous piece of utterance we may perceive certain
variations of prominence, characterizing its constituent sounds
in such a way that the more prominent alternate with the less
prominent in a more or less regular succession. A diagrammatic
representation of the opening of the Aeneid, for example, would
appear somewhat as follows in terms of relative prominence:

It will be seen that the heights of the 'peaks' and the depths of
the 'valleys' are various; but it is their relative and not their
absolute measurement that is important from the standpoint of
linguistic structure (omitted for present purposes is the heighten-
ing of certain peaks as a consequence of stress or intonation).
In the above example there appear seven peaks, with six valleys
between them, thus:
peaks a - a - i - u - e - a - o
valleys m-u-r-qu-c-n
The number ofsYLLABLEsina piece generally corresponds to
the number of peaks of prominence. The sounds which habitually
occur at these peaks we term VOWELS, and those which occur
in the valleys CONSONANTS.
The classification is not, however, entirely straightforward.
Thus the r of uirum is in the valley, but that of arma is on the
slope; the point here is that in uirum the r is less prominent than
both the preceding i and the following t/, whereas the r in arma,
whilst less prominent than the preceding <z, is more prominent
than the following m. Similar considerations apply to the m's
 PHONETIC INTRODUCTION

of arma and of uirumque. At this point it will be advisable to

consider the nature of the so far undefined 'prominence'. In
Daniel Jones' words, 'The prominence of sounds may be due
to inherent sonority (carrying power), to length or to stress or
to special intonation, or to combinations of these'; so far as
concerns the vowel/consonant distinction, inherent sonority is
the most generally relevant factor—but there are exceptions.
Thus the initial u of uirum lies in a valley, whereas the u of the
second syllable forms a peak; yet the articulation of both by
the tongue and lips is more or less identical. Here the point
is that in its position before i the initial u is reduced to a very
short duration,1 with consequent loss of prominence (although
its inherent sonority is comparable with that of the i); the u
of the second syllable, on the other hand, is of high sonority
and prominence in contrast with the surrounding r and m.
Similar principles would apply to the two fs in a word such
as iussit.
Sounds which may function either as peaks or as valleys of
prominence, whilst classified as vowels in their peak (or
'nuclear') function, are generally termed SEMIVOWELS, and
classed with the consonants, in their valley (or * marginal')
function. Thus Latin i and u may represent both vowels and
consonants, and Latin does not distinguish the two functions in
writing—unlike e.g. English, which distinguishes i and u from
y and w.
Finally, it may be noted that two vowels can follow one
another as independent peaks, by means of some diminution of
energy between them: thus e.g. Latin a-it,faci-at, abi-it, mortu-us,
medi-us, tenu-is.

(ii) Consonants
A primary classification of consonants is into the categories of
VOICED and VOICELESS. Voiced sounds involve an approxima-
tion of the two edges of the vocal cords, so that when air passes
through them it sets up a characteristic vibration, known tech-
nically as 'glottal tone' or V O I C E ; voiceless sounds involve a
1
Probably also with some relaxation of lip-rounding.
 CONSONANTS
clear separation of the cords, so that no such vibration occurs.
The difference may be exemplified by the English (voiced) z
and (voiceless) s. If the ears are closed, the vibration of the
former can be clearly heard by the speaker; the vibration can
also be felt by placing a finger on the protuberance of the
thyroid cartilage ('Adam's apple').
Sounds may be further classified according to the position or
organ involved in their articulation. Thus LABIAL (or BI-
LABIAL) involves the articulation of the two lips (e.g. English
/?), LABIO-DENTAL the articulation of the upper teeth and
lower lip (e.g. English / ) , DENTAL the articulation of the
tongue-tip and upper teeth (e.g. English th), ALVEOLAR the
articulation of the tongue-tip and upper gums (e.g. English /),
VELAR the articulation of the back of the tongue and the back
of the palate (e.g. English k).
If the tongue or lips form a complete closure, during which
air is prevented from passing through the mouth until the
closure is released, the resulting sound is termed a STOP. Stops
are further subdivided into PLOSIVES and NASALS according
to whether the nasal passages are closed or open during the
articulation of the stop; thus English has the plosives p> b
(bilabial, voiceless and voiced), t, d (alveolar), k> g (velar), and
the nasals m (bilabial), n (alveolar), and as ng in sing (velar).
In most languages the nasals are all inherently voiced.
The ' plosion' of the plosives refers to the effect which is pro-
duced when the oral closure is released. If the vocal cords are
left open for a brief period after the plosion, producing an
audible type of 'A-sound', the consonant is termed ASPIRATED
—there is clear aspiration, for example, of voiceless plosives at
the beginning of words in English and German. In French, on
the other hand, the vocal cords are closed almost simultaneously
with the oral plosion, and the result is a relatively UNASPIRATED
sound.
If the articulating organs are not completely closed, but if the
channel between them is so narrow as to cause an audible
effect as the air passes through it, the resulting sound is termed
a FRICATIVE. English examples are / and v (labio-dental,
 PHONETIC INTRODUCTION

voiceless and voiced), and s and z (alveolar). The A S P I R A T E ,

h, is also sometimes called a 'glottal fricative'.
If one side of the tongue forms a closure, but the other side
permits air to flow freely,1 the result is a L A T E R A L consonant,
such as the English /. Such sounds are sometimes classed with
the r-sounds as 'liquids' (see p. 32).

(iii) Vowels
Variations of vowel quality are effected primarily by the raising
of different portions of the tongue's surface towards the palate,
and by different degrees of such raising resulting in different
degrees of aperture between tongue and palate. Vowels may
thus be classified according to (a) how far FRONT or BACK they
are articulated (i.e. involving more forward or more backward
areas of the tongue and palate), and (b) how CLOSE or O P E N
they are (i.e. involving greater or lesser raising of the tongue).
The relations of the vowels to one another may then be con-
veniently represented in terms of a two-dimensional diagram.
When so represented they tend to fall into a triangular or
quadrilateral pattern,2 such as:
.FRONT BACK

Vowels intermediate between front and back are referred to as

CENTRAL, and vowels intermediate between close and open as
MID (the so-called 'neutral' vowel of standard southern British
English, as at the end of sofa orfinger,is a mid-central vowel).
1
Alternatively there may be a central closure, with air-flow on both sides.
2
It should be mentioned that such a pattern applies more exactly to the
acoustic effects of the vowels than to their actual physiological articulation.
 VOWELS
Associated with the features already mentioned are various
degrees of UP-ROUNDING; generally speaking back vowels are
associated with rounding and front vowels with its absence
(lip-spreading). Thus the English u and i, in e.g. put, pit, are
respectively close back rounded and close front unrounded.
Sometimes, however, rounding is associated with a front vowel
and spreading with a back vowel—thus the French u, German
ii, and classical Greek u are front rounded vowels, and back
unrounded vowels occur in some languages.
Vowels are normally articulated with the nasal passages
closed (by raising the soft palate or 'velum'), but if the nasal
passages are left open the result is a NASALIZED vowel (as e.g.
in French on, phonetically transcribed [5]).
D I P H T H O N G S are formed by articulating a vowel and then,
within the same syllable, making a gradual change of articula-
tion (or 'glide') in the direction of another vowel. Most com-
monly, but not inevitably, the first element of a diphthong is
more open than the second. Thus the diphthong of English high
involves a glide from a towards i, of how from a towards u, and
of hay from e towards i.
In many languages vowels fall into two degrees of LENGTH,
LONG and SHORT. By and large the difference corresponds to
a greater as opposed to a lesser duration—-but not invariably so.
Other features, such as muscular tension, difference of quality,
and tendency to diphthongization, may be at least as im-
portant (they are, for example, in distinguishing the so-called
' short' vowel of English bit from the so-called ' long' vowel of
beat). f

(iv) Accent +
A C C E N T is a general term covering two distinct linguistic func-
tions, and two different modes of implementing these functions.
The two functions of accent are termed ' delimitative' and
'culminative'. The first of these, as its name suggests, concerns
the fact that in certain languages there are restrictions on the
position of the accent within the word such that, given this
position, it is possible to infer from it the boundaries of words.
5
 PHONETIC INTRODUCTION

Thus in Czech or Hungarian, words are normally accented by

stress on their first syllable; the occurrence of stress in these
languages thus indicates the beginning of a word. In Armenian,
words are normally stressed on their final syllable, so that the
occurrence of stress here indicates the end of a word. In Polish,
words are normally stressed on their penultirr xe syllable, so
that the occurrence of stress indicates a word-boundary after
the next syllable. The accent of classical Latin is delimitative
in a rather complex manner (see p. 83), but that of the
majority of Greek dialects only trivially so. In English or
| Russian, for example, where the accent is free (cf. English
import, import; Russian muka 'torment', mukd 'flour'), the accent
is of course not delimitative, since it is impossible to predict
word-boundaries from it. In such cases the accent has only its
* culminative' function of indicating the number of full words
in the utterance1 (a function that is also included in the de-
limitative accent); the culminative function may in fact be
considered as a phonetic expression of the individuality of the
word, focused upon a particular portion of it.
Whether delimitative or merely culminative in function, two
specific modes of accentuation must be recognized, (a) PITCH,
or TONAL accent, and (b) STRESS, or DYNAMIC accent. The
tonal accent involves a raising of the voice-pitch at a particular
point, and the dynamic accent involves an increase in the
X muscular effort (primarily by the abdominal muscles).
It is important to distinguish tone from INTONATION. The
former refers to the pitch-patterns operative within individual
words, whereas 'intonation' refers to the pitch-pattern operative
over the whole clause or sentence. There may of course be, and
there usually is, considerable interaction between these two
patterns; thus the pitch-pattern of a given word may vary
greatly in accordance with the pitch-pattern of the sentence;
such an effect is sometimes referred to as a 'perturbation5 of the
word-tones. Rather similar considerations apply in the case of
stress, though one might expect the ' perturbation' to be less in
1
Since it is free, however, it is capable, unlike the delimitative accent, of
carrying differences of meaning (as in the English and Russian examples cited).
 ACCENT

those languages where the word-stress is strong, as, for example,

in English; even here, however, some variation is possible, as
for instance in the word fundamental in the two sentences it's
quite fundamental and it's a fundamental principle. In French, stress
is a feature of the word only as an isolate (in which case it falls
on the final syllable); in connected speech, however, it is rather
a feature of the sense-group.
Naturally the syllables in a word have varying degrees of
stress, but by the stressed syllable we mean the syllable which
carries the main stress.
Ceteris paribus, stressed sounds produce greater intensity of
air-pressure and are perceived as louder than others; but, as
already mentioned, the overall prominence of a sound depends
upon other features also, such as inherent sonority, duration
and intonation; and it is not always easy to disentangle the
various causes contributing to its perception.
The distribution of accentual types amongst the languages of
the world is a matter for observation rather than prediction.
Some more or less universal rules do, however, seem to be
emerging. For example, it has been claimed that if a language
has significant distinctions of vowel-length, as Latin or Greek,
it will not generally have a free dynamic^accent; and if a lan-
guage requires an analysis of its syllabic peaks into 'morae' (as
classical Greek), its accent is likely to be tonal, but if (as in
Latin) no such analysis is demanded, the accent is likely to be
dynamic.

(v) Speech and writing

In the study of a 'dead' language there is inevitably a main
emphasis on the written word. But it is well to remember that
writing is secondary to speech, and however much it may de-
viate from it, has speech as its ultimate basis. The written
symbols correspond, in a more or less complete manner, to
phonological or grammatical elements of speech; and, as Mar-
tinet points out, * vocal quality is directly responsible for the
linearity of speech and the consequent linearity of script'. It is
therefore in a sense misleading to speak of written symbols as
 PHONETIC INTRODUCTION
being pronounced—rather is it the other way round, the symbols
represent spoken elements. But when, as in the case of Latin
today, most utterance consists of reading from a written text,
the traditional terminology of' pronouncing letters' may reason-
ably be tolerated, and is in fact maintained in this book.
In Latin, as in modern European languages, the corre-
spondence is between symbols (letters) and phonological ele-
ments, and is much more regular than in some languages, such
as English or French or Modern Greek, which notoriously use
different symbols or combinations of symbols to indicate the
same sound.
It is sometimes stated that an ideal writing-system would
have a symbol for every sound—that it would in fact be a kind
of'visible speech'. Since, however, the number of sounds in a
language is infinite, and the ' same' sound probably never pre-
cisely recurs, this requirement is quite impracticable. It is also
unnecessary, as alphabets from earliest times have recognized.
The number of symbols can be reduced to manageable propor-
tions without any resultant ambiguity by a process which has
long been unconsciously followed, and the theoretical basis of
which has been worked out in recent years.
What is required is not one symbol per sound, but one
symbol per PHONEME. A* phoneme' is a class of similar sounds
that are significantly different from other sounds, e.g. the class
of /-sounds in English tin, hat, etc., or the class of (/-sounds in
din, had, etc. The (voiceless) /-phoneme and the (voiced)
(/-phoneme are different phonemes in English, and so require
distinct symbols, because tin has a different meaning from din,
hat has a different meaning from had, etc.; in technical termino-
logy, the members of the t and d phonemes are in * parallel
distribution', i.e. they can contrast significantly with one
another, and with members of other phonemes, in otherwise
identical immediate environments ((-)in, ha(-), etc.).
On the other hand, the fact that an initial t in English (as in
tin) is more strongly aspirated than a final / (as in hat) is not
responsible for any difference of meaning, since the two varieties
occur only in different environments, and so cannot contrast
8
 SPEECH AND WRITING
with one another—they are in 'complementary' and not
parallel distribution. They are thus both members (or 'allo-
phones') of the same /-phoneme; only one symbol is required
to write them, since the difference in sound is predictable from
their environment, i.e. initial or final position as the case may
be. It should be noted, however, that the phonemic distribution
of sounds varies from language to language; in a language such
as classical Greek or modern Hindi, for example, aspirated and
unaspirated /-sounds belong to separate phonemes, since the
occurrence of one or the other is not predictable from environ-
ment and they may contrast significantly (e.g. Greek TEIVCO
'stretch', Oeivco 'strike'; Hindi sat 'seven', sdth 'with').
The number of phonemes in a language varies; the number
of consonants, for example, varies from 8 in Hawaiian, through
24 in English and 32 in Sanskrit, to 80 in the Caucasian Ubykh.
Latin, according to the analysis adopted,1 has from 15 to 18
consonant phonemes in native words.
This 'phonemic' principle, then, is an economic principle,
ensuring that the minimal number of symbols are used con-
sistent with unambiguous representation of speech. And Latin
spelling comes very near to being completely phonemic. The
principal shortcoming in this respect concerns the vowels, since
no distinction is made in standard orthography between short
and long—thus, for example, malus 'bad' and uictum (from
uinco) are not distinguished from malus 'apple-tree' and uictum
(from uiuo); also no distinction is made between consonantal
and vocalic i and u, as in adiecit, adieus and inuitus, minuit, etc.
(uoluit provides a case of actual ambiguity).
When indicating particular sounds in a phonetic notation it
is customary to enclose the symbols in square brackets, e.g. [th]
to represent the initial sound of English tin; phonemic symbols,
on the other hand, are conventionally set between obliques,
e.g. /t/ for the phoneme which includes the initial sound of tin
and the final sound of hat. In a book intended primarily for the
1
Depending upon whether the [g] sounds ofmagnus, incipio, etc. (see pp. 23, 27)
are classed together as a separate phoneme, and whether qu> gu are treated as
single phonemes (though represented by digraphs) or as sequences of c, g and
consonantal u (see pp. 16, 25).
 PHONETIC INTRODUCTION
classical and general reader rather than the technical linguist
and phonetician it has seemed desirable to keep phonetic symbols
to a minimum. This inevitably involves some theoretical mixing
of phonetic, phonemic, and graphic levels of statement, but no
practical confusion should thereby be caused. A rigorous separa-
tion of levels (as is necessary on e.g. pp. 15, 28) would lead to
greater complexity of statement, which would tend to obscure
the primary purpose of this study. For the same reason the
conventions of the International Phonetic Alphabet have in some
cases been modified in the direction of more familiar forms—
e.g. by the use of [y] instead of [j] for the palatal semivowel (where
the latter could be misleading to the general English reader),
and by the use of the macron instead of the colon for vowel
length.
Note: Where English equivalents are given for Latin sounds,
the reference, unless otherwise stated, is to the standard or
'Received Pronunciation' (R.P.) of southern British English.
The choice of this form of English as a basis of comparison is
made on purely practical grounds. It is impossible to cite
examples that will be equally applicable to all nationalities and
dialects of English, and one must perforce select a standard;
and 'R.P.' is by far the best documented and familiar of such
standards. Nevertheless, care has been taken to select examples
which, so far as possible, will not be positively misleading to
speakers of other forms of English.

10
 CHAPTER I

CONSONANTS

Before considering the individual sounds in detail, it is important

to note that wherever a double consonant is written in Latin it
stands for a correspondingly lengthened sound. This is clearly
seen from its effect on the quantity of a preceding syllable,
the first syllable of e.g. accidit or Me always being 'heavy'
(see p. 89) although the vowel is short. Quite apart from
metrical considerations, it is necessary to observe this in pro-
nunciation, since otherwise no distinction will be made between
such pairs as ager and agger, anus and annus. English speakers
need to pay special attention to this point, since double con-
sonants are so pronounced in English only where they belong
to separate elements of a compound word—as in rat-tail, hop-
pole, bus-service, unnamed, etc.; otherwise the written double con-
sonants of English (e.g. in bitter, happy, running) have the function
only of indicating that the preceding vowel is short. The English
compounds in fact provide a useful model for the correct pro-
nunciation of the Latin double (or 'long') consonants.
In early systems of Latin spelling, double consonants were
written single; the double writing does not appear in inscriptions
until the beginning of the second century B.C. Ennius is said
to have introduced the new spelling (cf. Festus, under soli-
taurilia), but in an inscription of 117 B.C. the old spelling is still
more common than the new.1 The single spelling in such cases
does not of course indicate single pronunciation, any more than
the normal single writing of long vowels indicates a short
pronunciation.
1
Another device, mentioned by the grammarians and occasionally found in
Augustan inscriptions, is to place the sign 'sicilicum* over the letter to indicate
doubling (in the manner of the Arabic 'shadda')—thus, for example, o$A = ossa.

II
 CONSONANTS

(i) Voiceless* plosives*1

There are four varieties of these in Latin—bilabial*, dental*,
velar*, and labio-velar (see p. 16); they are written as/?, t9 c,
and qu respectively. The first three have a close affinity to the
sounds represented by English/?, t, k (or 'hard' c).
The English voiceless plosives, particularly at the beginning
of a word, are clearly aspirated*. The corresponding Latin
sounds were relatively unaspirated, as is shown by the fact that
they were generally transcribed as TT, T, K respectively in Greek
(e.g. KOOTETCOAIOV, KOIVTOS for Capitolium, Quintus); for the Greek
letters can only stand for unaspirated plosives. The Romance
languages also generally agree in lacking aspiration (e.g. the
pronunciation of Spanish tiempo, from Latin tempus).
But since, as in English and unlike Greek, there was no con-
trast in native Latin words between unaspirated and aspirated
plosives, and so no possibility of significant confusion, some
degree of aspiration could theoretically have been tolerated;
and one piece of evidence, though indirect, is rather suggestive
in this connexion.
When an English speaker listens to an Indian language such
as Hindi (which, like ancient Greek, distinguishes between
aspirated and unaspirated consonants), he tends to interpret
the unaspirated voiceless plosives, particularly at the beginning
of a word, as if they were voiced* (hearing/? as b, k as g, etc.).
The reason is that, since voicelessness in English is normally
associated with aspiration, complete absence of aspiration, as
f in Hindi /?, /, k, etc., is heard as voice*, for, without special
training, we inevitably listen to a foreign language in terms of
our native system of phonemes. Now there are some Greek
words containing initial voiceless unaspirated consonants which
are borrowed into Latin with voiced consonants; thus xufJcpvco
becomes guberno, TTU^OS becomes buxus, KOUUI becomes gummi,
KpA(3(P)crTOS becomes grab{b)atus> and so on; which could mean
that in this respect the Roman listening to Greek was in much
1
An asterisk after a term indicates that it is explained in the phonetic intro-
duction.

12
 VOICELESS PLOSIVES

the same situation as the Englishman listening to Hindi, i.e. that

the voiceless plosives of his own language, at least in initial
position, tended to be aspirated.1 A number of the words so
borrowed appear to have been of a colloquial character, and
they may be further augmented from Vulgar Latin, as e.g.
(reconstructed) botteca from orrro0f|Kr| (cf. Italian bottega), or
(Appendix Probi, K. iv, 199)2 blasta from TrActor-. That the
tendency was also prevalent in earlier times is evident from
Cicero's statement (Or. 160) that Ennius used always to say
'Burrus' for Pyrrhus. In fact the phenomenon seems to be parti-
cularly associated with non-classical borrowings, in which the
actual speech is likely to be reflected rather than a literary
consciousness of the Greek spelling.3
It is admittedly a minor detail that is in question; but the
discussion will have served to show how light may sometimes
be shed on ancient linguistic problems by the observation of
modern parallels; and on a more practical level, that one
should probably not insist too strongly on the complete
avoidance of aspiration in Latin.
t It is sometimes said that the Latin sound represented by this
letter differed from the comparable sound in English, since the
latter does not have a dental* but an alveolar* articulation
(in which the tongue makes contact with the gum-ridge behind
the upper teeth rather than with the teeth themselves); whereas
Latin, to judge from the evidence of the Romance languages,
had a true dental articulation (as, for example, in French). It
should, however, be mentioned that the grammarians do appear
to prescribe something not unlike the English alveolar contact,
in contrast with a pure dental contact for the voiced sound of d
(e.g. fTerentianus Maurus, K. vi, 331).4 It would be unwise
1
Though presumably less so than those which later came to be written as such
(see p. 26).
2
This and similar references are to the volumes of Keil's Grammatici Latini.
8
It has been pointed out that many of the words in question are probably
non-Indo-European, and borrowed by both Greek and Latin independently from
some * Mediterranean' source. But this does not invalidate the argument, since
the different forms in which they were borrowed indicate a different interpretation
of the sounds by Greek and Latin speakers.
4
Texts of references marked thus (f) are given in Appendix A (1).

13
 CONSONANTS

to make too much of this evidence; for the Greek and Latin
grammarians never succeeded in discovering the general dis-
tinction between voice and voicelessness,1 and so were quite
liable to seize on any minor, or even imaginary, difference of
articulation in order to distinguish between a particular pair of
sounds (cf. p. 21). But at the same time the existence of such
statements once again makes it questionable whether one should
insist on suppressing English speech-habits in this particular
connexion.

c Latin c in all cases represents a velar* plosive—i.e., in popular

terminology, it is always 'hard* and never 'soft'—even before
the front* vowels e and i. Inscriptions in fact sometimes write
k for c in this environment (e.g. pake), and Greek regularly
transcribes Latin c by K (e.g. KT|vacop, KiKepcov for censory Cicero);
the sound was also preserved in words borrowed from Latin by
Celtic and Germanic between the first and fifth centuries A.D.
In the grammarians there is no suggestion of anything other
than a velar plosive; and Varro (Priscian, K. ii, 30) provides
positive evidence by citing anceps beside ancora as an example
of the velar value of n (see p. 27)—which only makes sense if
the following sound is the same in both cases. There is a further
hint in the alliterative formula 'censuit consensit consciuiV (Livy,
h 32, 13)-
It is true that in the course of time a 'softening' took place
before e and i (compare the pronunciation of c in French cent,
Italian cento, Spanish ciento, from Latin centum); but there is no
evidence for this before the fifth century A.D. ; and even today
the word for ' 100' is pronounced kentu in the Logudoro dialect
of Sardinia.
This of course does not mean to say that Latin c represents
an absolutely identical pronunciation in all environments. In
1
It was, however, already familiar to the earliest of the Indian grammarians
and phoneticians (Allen, Phonetics in Ancient India, 33 ff.). Quintilian (i, 4, 16)
recommends the learning of the tjd distinction, but does not discuss it. In the
middle ages the grammarian Hugutio still admits: 'licet enim d et t sint diuersae
litterae, habent tamen adeo affinem sonum, quod ex sono non posset perpendi
aliqua differentia'. Not until the nineteenth century is the distinction clearly
understood in Europe.

14
 VOICELESS PLOSIVES

English, for example, the initial sound in kit is articulated

somewhat further forward on the palate than in cat, and is
accompanied by a certain degree of lip-rounding* in coot.
There is perhaps some actual evidence for this in Latin; an
original short e followed by a 'dark' / (see p. 33) normally
developed to a back* vowel, 0 or u—thus Old Latin helus
becomes holus, and the past participle ofpello is pulsus; but scelus
does not change to scolus, and the past participle of the obsolete
cello is celsusy not culsus; one possible explanation of this is that the
change was prevented by the frontness of the preceding consonant, f
In early Latin inscriptions c tends only to be used before
i and e, k before consonants and a (retained in Kalendae and in
the abbreviation K. for Kaeso), and q before 0 and u—e.g. citra,
feced; liktor, kaput; qomes, qura—which is a further indication that
the pronunciation varied somewhat according to environment;
this practice is also found in some early Etruscan inscriptions.
Such a complication, however, was clearly unnecessary; it is
'unphonemic' (see pp. 7ff.)and would involve, if consistently
employed, such variations as loqusy loka, loci within a single
paradigm; and c was subsequently generalized in all environ-
ments, except in the consonantal combination qu.
The letter-shape c was ultimately derived from the Greek
gamma ( f ) , through a stage <; but, as we have seen, it had
come to be used in early Latin writing as a positional variant
with K and Q, (which it later supplanted) as a sign of the voiceless
velar plosive /k/. This meant that there was no longer any
distinctive sign for the voiced /g/ (hence inscriptional forms such
as v i R c o for uirgo). In Etruscan, which perhaps provided the
model for the Latin practice, this did not matter, since in that
language voiced and voiceless plosives seem not to have been
significantly distinguished. But in Latin the voiced /g/ con-
trasted with the voiceless /k/ (e.g. lugere: lucere); and the dis-
tinction between the two phonemes was eventually indicated
by introducing the symbol G for the voiced consonant (formed
perhaps by the addition of a stroke to c). 1 The old spelling is,
1
The device is traditionally attributed to Sp. Carvilius Ruga (third century
B.C.), but it may go back to Appius Claudius in the late fourth century.

15
 CONSONANTS

however, preserved in the abbreviations C. for Gains and Cn. for

Gnaeus.

qu The sound represented by this ' digraph' was of a type

known technically as LABIO-VELAR, i.e. a velar plosive (such
as that represented by Latin c) but with a simultaneous
rounding and protrusion of the lips (as for English w); the
phonetic symbol for such an articulation is [kw].
It is fairly certain that it was not a matter of two successive
consonants as in e.g. English quick, where qu represents [kw];
for this we have some evidence in the grammarians, who speak
of the ^-element as being part of or blended (confusa) with
the preceding letter (Pompeius, K. v, 104; Velius Longus,
K. vii, 58; cf. Ter. Scaurus, K. vii, 16).1 A statement of Marius
Victorinus, though not altogether clear, seems in fact to dis-
tinguish the sound of c or k from that of qu simply by openness
versus protrusion of the lips (|K. vi, 34).
The grammarians' statements are supported by the fact
that, with very rare exceptions, qu does not cmake position5
in verse as it might be expected to, at least optionally, if it re-
presented a sequence of two consonants; thus the first syllable
of e.g. equi is always light. However, against this it could
be argued that the treatment of certain groups as alternatively
'making position' is borrowed from Greek (see pp. 8gf.), and
that, unlike the groups plosive + liquid* (tr> etc.), a group [kw]
had no parallel in Greek, which had lost its w at an early
date.
Another fact which is sometimes cited as proof of the simul-
taneous nature of the ^-element is that an m before qu may
remain unchanged, whereas before c it is regularly changed to
n (= [Q] ; see p. 27); thus horum + ce gives horuncy am + ceps gives
anceps, but quam + quam remains quamquam (similarly quicumque,
numquam, umquam, etc.)—which suggests that the labial w-
element was present from the start of the yw-sound, thereby
providing an environment that favoured the preservation of the
1
Quintilian's apparent citation of quos (i, 4, 10) as an example of consonantal
u is probably a wrong reading (cf. Coleman, CQ,, N.S. XIII (1963), 1).

16
 VOICELESS PLOSIVES
preceding labial m (just as in e.g. quamuis),1 in spite of the
velar articulation of the stop* element. But the existence of
alternative spellings with n (quanquam, etc.),2 and the possibility
that the m is due to analogical influence from quam, cum, etc.
(as in e.g. quamdiu beside inscr. quandiu) diminishes the value of
this evidence.
There are also historical arguments of two types. First, it is
notable that, whereas in other environments the consonantal u
[w] eventually became a fricative* [v], this change did not
affect the u of qu (thus Italian vero but quanto); a difference is
in fact already noticed by Velius Longus in the second century
A.D. (f K. vii, 58). Such a variation in development could of
course be attributed simply to the fact that in qu the u occurs
after a syllable-initial plosive, which is not the case in other
occurrences of consonantal w3 (except for gu; see below). But
the very fact that no other such groups occur (i.e. no syllable-
initial p, b, ty d 4- consonantal u) could itself be interpreted as
an indication of the special nature of qu (and gu).A One may
further cite here the statement of Priscian ( | K . ii, 7) that the
u element of qu, when followed by a front vowel, has a special
quality, like the Greek u (i.e. like the initial sound of French
huit as contrasted with oui)—whereas this is not stated to apply
in the case of the independent consonant w.5
The other historical argument relates to the fact that in
nearly all cases Latin qu derives from a single, labio-velar
consonant of Indo-European, which is represented by single
consonants of various kinds in other languages; thus Indo-
1
Also in e.g. inscr. comualem (117 B.C.), which preserves the old prefix com-
(later con-).
2
Favoured by Pliny the Elder, according to Priscian (K. ii, 29).
3
In compounds such as aduenio, subuenio, the syllabic division falls between the
d or b and the consonantal w, giving regularly heavy quantity to the preceding
syllable (cf. p. 89).
4
It was already so interpreted by the grammarians Pompeius and Ps.-Sergius
in their commentaries on Donatus (K. iv, 367; iv, 476; v, 104), though their
arguments are misunderstood by Bede (K. vii, 228).
6
This is confirmed also for the classical period by Greek inscriptional spellings
such as KuivnXios, AKUA(A)IOS for Quintilius, Aquilius (Augustus or earlier), with KUI
or KU for qui, as against KOVCC or KOOC regularly for qua. No such spelling is found for
simple ui [wij; cf. Eckinger, Die Orthographie lateinischer Worter in griechischen
Inschriften. See also p. 52 below.

17
 CONSONANTS
w
European k od gives, beside Latin quod, Sanskrit had: Oscan
pud: English what (where wh is pronounced either [hw] or simply
[w]). However, this argument is slightly weakened by the fact
that in equus the qu derives from an Indo-European group kw,
which is represented by a group or double consonant in some
other languages (e.g. Sanskrit asvas: Greek ITTTTOS).
The various arguments, at least on balance, clearly favour
the pronunciation of qu as a single, labio-velar consonant [kw].
Consonants of this type are common in a number of languages,
e.g. at the present day in Caucasian, African, and American
Indian languages, and in ancient times in Mycenaean Greek;
and they present no particular difficulty of pronunciation. On
the other hand no confusion is caused if the Latin qu is pro-
nounced in the same way as English, since a sequence of [k]
and [w] does not otherwise occur in Latin (though it does in
some of the languages mentioned above, including Mycenaean).
It is possible in any case that an alternative pronunciation of this
kind may actually have existed in some varieties of Latin speech.
In Lucretius, for example, some occurrences of the forms aquae
and dquai are probably to be read with heavy first syllable;
this is certainly true of liquidus in some cases;1 and such treat-
ment becomes more common in Latin poetry after the fourth
century A.D. This might indicate a pronunciation of qu as a
t consonant-group [kw]—as is assumed for the Lucretian examples
by the grammarian Audax (K. vii, 328 f.); on the other hand,
it may simply reflect the beginning of a dialectal tendency to
lengthen the stop element to [kk]—a tendency reproved in the
fourth century by the Appendix Probi (K. iv, 198; 'aqua, non
acqua'), though with little effect to judge by Italian acqua.
There remains one further peculiarity connected with qu, the
discussion of which requires a certain amount of preliminary
explanation. Under various conditions, and probably at various
times, Old Latin short 0 became classical u (e.g. during the
third century B.C. in final syllables, so that primos became primus,
etc.). Where, however, the 6 was preceded by u (vowel or
consonant), or by qu or ngu, these changes do not appear in
1
(for aqua) vi, 552, 868, 1072; (for liquidus) i, 349; iii, 427; iv, 1259.

18
 VOICELESS PLOSIVES
writing until the end of the republic. Until then inscriptions
still show such forms as uolgus, auonculus, seruos> perspicuos, equos,
instead of uulgus, etc. Scholars are divided in opinion as to
whether the spelling with uo really represents the pronunciation,
or whether it was preserved merely to avoid the ambiguous
writing of two successive u symbols,1 which might possibly be
interpreted as a single long vowel2 (cf. p. 64). The latter
explanation may not^ appear altogether convincing, but the
practice does seem to have an orthographic rather than a
genuine phonetic basis; for when the change of 0 to u does
eventually take place in the spelling of such words, it affects all
cases equally, whatever the phonetic conditions; thus words of
the type seruos (final syllable), uolgus (before / + consonant),
auonculus (before [Q]), all equally start to appear as seruus,
uulgus, auunculus, etc., in inscriptions of the Augustan period.
The old spelling is found not only in inscriptions, but also in
some manuscript traditions, as of Plautus and Terence, and
even Vergil and Horace. But wherever uo is found for later uu
in classical Latin, it is certainly to be pronounced as uu in
imperial times, and almost certainly earlier. This does not of
course apply to those cases where uo is invariable; thus, although
uolt is to be read as uult> uolo is to be pronounced as written.
In the case of Old Latin quo and nguo> however (as in e.g.
equosy unguont), a further development is involved; for when
the change to quu, nguu took place, the new u vowel had the
effect of causing a dissimilatory loss of the preceding u ele-
ment: thus quu, nguu became cu, ngu.z This phonetic change
1
The writing with o, however, does of course involve a secondary ambiguity,
since if vowel-length is not marked, seruos could stand for nominative singular or
accusative plural.
2
Thus Quintilian i, 4, 10; cf. i, 7, 26 and Velius Longus, K. vii, 58 f. Con-
versely, towards the end of the republic, uu came to be written where a single u
would give rise to ambiguity: thus, for example, iuuenisffluuius, as against earlier
inscr. iuenta, Jluio, etc., where the i might be read wrongly as a vowel in the first
word and a consonant in the second (no such ambiguity arises, however, in a form
such asjluit, which therefore continues to be so written).
3
A similar loss of consonantal u before u vowels probably occurred in other
cases also: thus, beginning at the end of the republic, occasional inscriptional
forms such as aeum, uius, serus for aeuum, muus, seruus. But in such cases analogical
pressure (from serui, etc.) rapidly restored the lost u both in writing and in educated

19
 CONSONANTS
is largely obscured by analogical spelling (e.g. equos or equus
continuing to be written for ecus after the analogy of such forms
as the plural equi); but the true situation is revealed by occa-
sional inscriptional forms with c, and later confirmed by the
grammarians, who, though they support the analogical spellings,
are nevertheless clear that they do not correspond to the pro-
nunciation. Thus (first century A.D.) Cornutus (Cassiodor(i)us,
K. vii, 150 f.): 'extinguunt per duo u.. .extinguo est enim, et
ab hoc extinguunt, licet enuntiari non possit'; (second century
A.D.) Velius Longus (K. vii, 59): 'auribus quidem sufficiebat
ut equus per unum u scriberetur, ratio tamen duo exigit5.
This dissimilatory loss of u may well have been an immediate
consequence of the change of 6 to u; so that in classical Latin
wherever one finds qua or nguu (or quo, nguo in the older spelling),
they probably represent a pronunciation cu, ngu. Thus equus or
equos probably stands for ecus, quum or quom for cum, sequuntur
or sequontur for secuntur, unguunt or unguont for ungunt, etc. But no
doubt there were analogical pronunciations, as well as spellings,
of the type equus, and such a pronunciation is also therefore
probably admissible.1
No problem of course arises in the case of words like quod,
sequor, where there is never any change of 6 to u, and which are
therefore always to be pronounced as written.2

(ii) Voiced* plosives

There are four varieties of these in Latin, parallel to the voiceless
series, and represented by b, d, g, and gu respectively. The
speech (except in bourn, genitive plural of bos, which became normal). The forms
without consonantal u evidently survived, however, in some forms of popular
speech (cf. Appendix Probi, 'riuus non rius\ and Italian rio).
1
There remains the possibility that before an K-vowel c and g were in any case
pronounced as [k w ], [g w ], with an automatic 10-element. In which case it is not
so much a matter of dissimilation as of * neutralization', i.e. absence of difference
between cujngu and quujnguu.
1
The pronunciation of quoque ('also') is sometimes questioned. Quintilian
reports, as a pun in bad taste, Cicero's words to a candidate whose father was a
cook: 'Ego quoque tibi fauebo' (vi, 3, 47), which seems to suggest that quoque was
pronounced coque (similarly in Anthol. 199, 96). Etymologically this is a possible
form (cf. cottidie), but it is not otherwise attested, and other interpretations are
possible.

20
 VOICED PLOSIVES
grammarians, as we have seen in the case of the t/d pair (p. 13),
failed to discover the nature of' voice'; thus d is said to differ
from / in that it represents a pure dental. It may well in fact
be true that d was pronounced as a dental, but we can no more
safely rely on the grammarians' statements of this than we can
on their descriptions of f^as an alveolar. The difference between
b and />, and between g and c, is expressed in terms that could
be interpreted as referring to a difference in muscular tension,
which commonly supplements the voice difference (e.g. f Marius
Victorinus, K. vi, 33). But in some cases the writer is clearly
at a loss to explain the nature of the distinction—thus Martianus
Capella (3, 261): '2? labris per spiritus impetum reclusis edi-
camus... .P labris spiritus erumpit'.
b, d and g> then, haj/e close affinities to the voiced sounds
represented by English b, d and ' hard' g.
In some cases, however, b is written instead of p for the
voiceless plosive—namely before the voiceless sounds of / and s
under certain special conditions. It is so used when the voice-
less sound occurs at the end of a preposition or noun-stem
which, in other environments, ends in a voiced b. Thus in e.g.
obtineo, obsideo, subsideo, absoluo, trabs, urbs, plebs, caelebs, the b is
in fact partially assimilated to the following t or s, becoming
voiceless [p]; but it continues to be written with a b by analogy
with forms such as obeo, urbis, etc. (similarly the preposition abs
owes its writing with b rather than p to the alternative form
ab); in inscriptions one even occasionally finds such forms as
scribtura (with b after scribo).
On general phonetic grounds it is highly probable that the
b before t or s should stand for [p]. It is moreover expressly
stated by Quintilian (i, 7, 7) and other grammarians, and
clearly indicated by inscriptional spellings with p at all periods
(e.g. pleps, opsides, apsoluere, suptilissima, optinebit). The distinction
between spelling and speech is clearly summed up by Quin-
tilian in the words: 6b litteram ratio poscit, aures magis audrunt
p9; and on the writing of abs Velius Longus comments (K. vii,
62): *qui originem uerborum propriam respiciunt, per b
scribunt'.
21
 CONSONANTS
Similar considerations apply to the spellings bf (obfero, etc.),
bm (submoueOy etc.), bg (obgero, etc.), be (subcingo, etc.), and br in
the case of the preposition sub (subripio, etc.), though here the
assimilation is complete, giving a pronunciation offero, summoueo,
oggero, succingo, surripio, etc.
Similarly also analogical spelling with d is found in the case of
the preposition ad. It is fairly certain that in most cases the d
was completely assimilated to the following consonant in speech
(except //, iy u, or m); so that spellings of the type adsequor,
adtineo, adripio, adfui, adpono, adgredior, adcurro were probably
pronounced as assequor, attineo, arripio, affui, appono, aggredior,
accurro, etc. Apart from the existence of these latter spellings
alongside the analogical adsequor, etc., one may cite the pun in
Plautus (Po. 279):
M. Assum apud te eccum.
A. At ego elixus sis uolo,
which involves a play on assum (ad-sum) ' I am present' and
assum 'roasted'. The question of such spellings is raised by
Lucilius (375 Marx), though he dismisses it as unimportant:
. . . accurrere scribas
d-nc an c non est quod quaeras....
As Velius Longus (loc. cit.) comments, 'ille quidem non
putauit interesse scripturae; sed si sonus consulitur, interest
aurium ut c potius quam d scribatur'.
In the case of dl, however (e.g. adloqui), the same grammarian
does permit a pronunciation as such, as well as the assimilated
alloqui. It is in fact uncertain to what extent in educated speech
the analogical spellings may also have been reflected in pro-
nunciation. What is virtually certain is that, even when this
happened, the b or d will have been devoiced to [p] or [t]
before a following voiceless sound—so that the analogical pro-
nunciation would actually be of the type opfero, supcingo,
atsequor, atfui, atpono, atcurro, rather than obfero, etc.

g As in the case of c, this never implies a 'soft' pronunciation.

The evidence is parallel to that for c (e.g. Greek reAAios =
22
 VOICED PLOSIVES
Gellius; ingerunt cited as an example of the velar nasal); there
is no evidence for any change before e and i until around
500 A.D. As with c, however, some slight variation is probable
according to the following vowel (as in English gear, guard,
gourd); the fact that gelu does not become golu may indicate a
fronted pronunciation before e (see p. 15).
In one particular environment, however, g seems to have had
a markedly different value. In the position before the dental
nasal* n (e.g. agnus, dignus, regnum) it is probable that it repre-
sented a velar nasal sound [Q], like that of ng in English hang or
n in bank: so that gn in a word like agnus would be pronounced
like the ngn of an English word like hangnail.
This would be in line with a general tendency of Latin to
nasalize plosives before n (note e.g. Latin somnus = Sanskrit
svapnas, with Latin change of p to m before ri).1 It is also
indicated by inscriptional spellings such as ingnes, ingnominiae for
ignes, ignominiae. It would further explain why an n appears to
be lost in such forms as ignosco ( = in+gnosco) or cognatus
( = con +gnatus); for
(a) before velar sounds we know from the grammarians that
n represented a velar [Q] (see p. 27);
(b) ifgn in fact represents [on], then a combination con +gnatus
would theoretically imply a pronunciation [koggnatus];
but before another consonant the double [00] would then be
simplified to [0], giving [kognatus] (N.B. inscr. congnatus); and
such a pronunciation would be represented by a spelling
cognatus.
Further evidence comes from words of the type dignus,
lignum, ilignus. The words from which these are derived—decet,
lego, ilex—all have a short e vowel, and it is necessary to explain
the change of e to 1. Now such a change does regularly take
place before the sound [g]; beside Greek nrfyyco [teggd], for
example, the cognate Latin verb is tinguo; and whereas
con + scando gives conscendo, con + tango gives contingo (there is a
parallel to this in the Middle English change of [c] to [i] in such
words as England). The change of vowel in dignus, etc., there-
1
Cf. also inscr. amnegauerit for dbmgauerit.

23
 CONSONANTS
fore, is explained if g had here the value [rj]. The absence of
any such change in words like regnum, segnis is due to the fact
that the vowel is here long (cf. Latin rex and Greek TJKOC re-
spectively), and so is not affected by the change.
Of little primary value, but of interest as confirmatory
evidence, is the play on words in Plautus (Ru. 767) between
ignem magnum and inhumanum, and in Cicero {Rep. iv, 6) between
ignominia and in nomine—both of which at least suggest a nasal
value for g.
However, the awkward fact still remains that the develop-
ments in most Romance languages are better explained by
assuming the normal [g] rather than the nasal value for g in
the group gn. Thus lignum develops in exactly the same way as
nigrum in e.g. Old French lein/neir, Engadine lain/nair, S. Italian
liunujniuru (the g in each case having undergone a change to
i or u before the following consonant). An important exception,
nevertheless, is the conservative Sardinian, with e.g. linna, mannu,
konnadu from Latin ligna, magnum, cognatum (cf. Latin inscr.
sinnu for signum).
The grammarians also are strangely silent about any nasal
pronunciation of g, and in initial position Terentianus Maurus
seems to suggest the normal [g] value in the name Gnaeus when,
referring to the spelling of the name with Cn.y he says
g tamen sonabit illic quando Gnaeum enuntio
(K. vi, 351).

But in fact by this time any pronunciation of the initial G

must have been artificial; as Varro already observes (fr. 330
Funaioli), 'qui G littera in hoc praenomine utuntur, antiqui-
tatem sequi uidentur'. Varro also notes a spelling Naeum (and
f Naios is common in Greek).
A solution to the apparent contradiction of evidence for the
pronunciation of^n was proposed by C. D. Buck, who suggested
that the nasal pronunciation of g as [Q] was in fact the normal
development, but that subsequently a ' spelling pronunciation'
was introduced, first in educated circles, then more generally,
whereby g was given its more common [g] value. At what
24
 VOICED PLOSIVES
period such a change took place it is impossible to say, but
Sardinian suggests that it was very late, and for the classical
period the nasal pronunciation remains the more probable.
What must be emphasized, however, is that at no period of
Latin was gn pronounced as a 'palatal' [n], as in modern
Italian or French and as in the national pronunciation of Latin
by speakers of these languages.
A note on the pronunciation of Latin gn in England will be
found at the end of Appendix B.
It is generally assumed that g did not have a nasal pronuncia-
tion before m, as in legmen, segmentum, since the change of e to i
does not occur in these words. However, since original gm
seems to have given mm ([Link] fromflag-ma,[Link]),all
examples of gm may have arisen later, e.g. by syncope, after the
change of e to i was operative (cf. the unsyncopated forms
tegimen, integumentum). The possibility of a pronunciation of g as
[Q] here is therefore not entirely excluded—though it cannot
be safely recommended.

gu On the grounds of its graphic parallelism with qu, we

might expect that Latin gu (with consonantal u) also represents
a single, labio-velar consonant [gw], rather than a sequence
[gw]. It is, however, less easy to demonstrate this, since the
grammarians do not specifically discuss the matter, and the
combination occurs only after n (as in lingua), where the pre-
ceding syllable is in any case heavy and so can give no clue.
However, in view of the fact that all other plosive consonants
in Latin occur in pairs, voiceless and voiced {p/b, etc.), it is to
be expected that the voiceless [kw] would have a voiced counter-
part [g w ]; and there seems in fact to be an indirect indication
of this parallelism in a passage of Priscian already referred to
(p. 17 above). For after mentioning a special quality of the
u element of qu when followed by a front vowel, he goes on
specifically to say that the same applies to the u element of gu.
 CONSONANTS

(iii) Aspirates*
The digraphs ph, th, ch represented aspirated voiceless plosives—
not unlike the initial sounds of pot, top, cot respectively (see
p. 12). They occupy a peculiar place in the orthographic
system, since they are not found in the earliest inscriptions and
make their appearance only about the middle of the second
century B.C. They are then used, and become standard,
primarily in transcribing Greek names and loan-words con-
taining aspirated plosives (9, 0, x)> e-g« Philippus, philtrum,
Corinthus, cithara, thesaurus, Achaea, bacchanal, machina, chorus; and
in such cases it is likely that educated Roman speakers in fact
reproduced the Greek aspirates with more or less fidelity.
Before this time the Greek aspirates had been transcribed in
Latin by simple/, /, c (e.g. inscr. Pilemo, Corinto, Antioco), and
this spelling remains normal in some early borrowings from
Greek (e.g. purpura = Trop<pupcc, tus = 6uos, calx = x ^ l € ) But sub-
sequently (beginning in fact, on inscriptional evidence, already
by the end of the second century B.C.) aspirates made their
appearance in a number of native Latin words (and loan-words
without an original aspirate): thus in pulcher, lachrima, sepul-
chrum, bracchium, triumphus, Gracchus (also in the place-name
Carthago), less generally in lurcho, anchora, Orchus, and occa-
sionally in inscriptional forms such as chorona, centhurio, praecho,
archa, trichlinium, exerchitator,fulchra, Olymphi, Volchanus, Marchus,
Calphurnius—note also the Greek spelling Z0A91K10S (from early
first century A.D.), as well as T"Io(u)AxeP (first in mid first century
B.C. and frequent later).
We know from a statement of Cicero ( | Or. 160) that in his
time an actual change in the pronunciation of many such words
was taking place, and he himself came to accept pulcher,
triumphus, Carthago, though rejecting sepulchrum, chorona, lachrima,
Orchiuius. The grammarians show a good deal of disagreement
(e.g. fMar. Viet., K. vi, 21; Ter. Scaurus, K. vii, 20), and it
would be easy to dismiss the aspirated pronunciation as a mere
fashionable misapplication of Greek speech-habits. That such
tendencies did in fact exist we know from Catullus' poem about
26
 ASPIRATES
Arrius, with his pronunciation oicommoda as chommoda, etc. But
it is remarkable that in nearly all the attested cases the aspira-
tion occurs in the vicinity of a 'liquid' consonant (r or I).1 It
seems more probable, therefore, that the aspiration represents
a special but natural environmental development in Latin
itself,2 which may possibly have varied in different areas and
social strata. Had the digraphs not been introduced to represent
the Greek aspirates in the first place, Latin would have had no
need to indicate the aspiration ofpulc(h)er, etc., in writing, since
it was merely an automatic variant of the normal voiceless stops
(just as we do not need to indicate the aspiration of initial
voiceless stops in English). But once the digraphs had been
introduced in order more accurately to represent the pro-
nunciation of loan-words from Greek, it would be natural
enough to employ them also for writing similar sounds in Latin.
The practical outcome of these discussions is as follows. An
English pronunciation of Latin p, t, c, though not intolerable,
will certainly be rather more aspirated than the Latin. And
some special effort is therefore required in pronouncing the
aspirates ph, th, ch, if these are to sound distinct from/?, t, c. It
should perhaps also be emphasized that there is no justification
for pronouncing the aspirates as fricatives*—i.e. as in photo,
thick, loch; this is admittedly the value of 9, 0, x in Late Greek,
but it had not yet developed by classical Latin times.

(iv) Nasals*
n Most commonly this stands for a dental (or alveolar) nasal
sound [n], similar to the n in English net or tent, e.g. in nego,
bonus, ante, inde.
Before a velar or a labio-velar, however (as in uncus, ingens,
relinquo, lingua), it stands for a velar nasal [q] (as in English
uncle or anger). Quite apart from the general expectation that
it would be assimilated in this way, there is clear evidence in
1
For a more technical discussion of similar effects elsewhere in the history of
Latin and some other languages cf. Archivum Linguisticum, x (1958), 110 ff.
2
Aspiration in some proper names, e.g. Cethegus, Othot Matho, may perhaps be
of Etruscan origin (Cicero accepts aspiration only for the first of these).

27
 CONSONANTS
ancient descriptions, the earliest of which goes back to Accius
(second century B.C.), who wished to follow Greek practice by
writing e.g. aggulus, agcora for angulus, ancora (|Varro, cited by
Priscian, K. ii, 30)} In the first century B.C. Nigidius Figulus
not ineptly described the sound as 'intermediate between n
and g9 (tGellius, xix, 14, 7).
The same [Q] sound almost certainly occurred when the
preposition in was followed by a word beginning with a velar
or labio-velar (e.g. in causa). Similarly, when followed by a
labial (p, b, m) it was pronounced m, as shown by inscriptional
im pace, im balneum, im muro.
In words like consul, where the n is followed by the fricative s,
one would certainly not be wrong in pronouncing it normally;
but other pronunciations were current even among educated
persons in classical times. At a very early period n in such an
environment had lost its consonantal value (a common de-
velopment in many languages) and had been replaced by a
mere nasalization of the preceding vowel, which was at the
same time lengthened by way of compensation for the lost
consonant. Thus consol, censor became cosol, cesor. As a result,
in the earliest inscriptions one often finds spellings of the type
cosol (whence the archaistic abbreviation cos.), cesor, cosentiont,
etc.,2 alongside the spellings with n. In popular speech the
nasalization was eventually lost, and we are told that even
Cicero used to pronounce some such words without an n, e.g.
foresia, Megalesia, hortesia (Velius Longus, K. vii, 79). In Vulgar
Latin it must have been completely lost, for there is no sign of

"j" * By one widely held view of phonemic theory, the [Q] would have to be con-
sidered as belonging to a separate phoneme from [n], since the two sounds occur
in contrast in annusjagnus (see p. 23). But one can also take the view that in e.g.
ancora the [Q] is an allophone of/n/ and in agnus an allophone of /g/. Such an inter-
pretation is reflected in the Latin orthography; the Greek practice, however
(e.g. in ayyos, Syxvpa), identifies the [g] before a velar with the [Q] which occurs
before a nasal in e.g. -n-payua and is there interpreted as an allophone of the /g/
phoneme—hence the spelling with yin both cases. The Greek practice is ambiguous
only in the case of Syyovos, where the first y in fact has the value [g], and which no
doubt for this reason is generally written Iicyovos. In Latin such a practice, as advo-
cated by Accius, would be made ambiguous by the existence of such words as agger.
2
Note also Greek transcriptions such as 'OpTTiaios, KAT^S, Kqaopivos for
HortensiuSy Clemens, Censorinus.

28
 NASALS
it whatever in the derived words in the Romance languages
(e.g. Italian mese> sposa from me(n)sis, spo{n)sa). f
But in the official orthography the n was preserved or
restored, and this had its effect on most educated speech of the
classical period. Probably few speakers, however, were entirely
consistent, and their inconsistencies provided a happy hunting
ground for later grammarians; in Caper, for example (K. vii,
95), we find the quite artificial rule: 'omnia adverbia numeri
sine n scribenda sunt, ut milies, centies, decies; quotiens, totiens per
n scribenda sunt'. One is reminded of the 'rules' about the use
of shall and will in English (while these were being crystallized
by the grammarians between the seventeenth and early nine-
teenth century, the actual usage was erratic—in fact even in
written English will was about twice as common as shall in the
first person!). The only safe practical rule for the modern reader
in regard to Latin ns is to pronounce the n wherever it is written.
The same considerations apply to cases where n is followed
by the other Latin fricative, / ; hence inscriptional forms such
as cofeciy iferos for confeci, inferos. In the classical forms with
restored n, however, the n here probably stands not for a dental
or alveolar nasal but rather for a labio-dental*, formed by contact
of the lower lip and upper teeth in the same way as the fol-
lowing / . Pronunciations of this kind are common for some
English speakers in words like comfort, information. In Latin the
variation in republican inscriptions between n and m in such
cases (e.g. infimo, infectei, confice, beside eimferis, comfluont, and
even im fronte) clearly points to such a pronunciation; and
although the spelling with n was later generalized, the labio-
dental pronunciation probably continued.
Wherever the nasal consonant was pronounced before s or/,
it is certainly to be considered as a more or less artificial
restoration, rather than a natural retention. For, as already
mentioned, when the n was lost it gave rise to a lengthening of
the preceding vowel; but the classical pronunciation WITH »
also has a long vowel (for evidence see p. 65), which shows
that the n must first have been lost, and subsequently restored.
The development in a word such as consul, therefore, is:
29
 CONSONANTS
prehistoric consol; early Latin cosol; classical colloquial cosul;
classical literary consul.
As might be expected, the difference between popular speech
and official spelling in this matter gave rise to occasional
spellings in which n was introduced where it had never in fact
been spoken, e.g. thensaurus ( = Greek dTjcravpos), occansio, Her-
culens, all of which are specifically proscribed by the gram-
marians. Such spellings may of course in turn have led to
occasional pronunciations based upon them.

m At the beginning and in the interior of words the sound

represented by m presents no problem. It stands for a bila-
bial nasal, as in e.g. English mat or camp. There are, however,
points to notice where it occurs at the ends of words. In
general it seems to have been reduced (like the n before a
fricative internally) to a mere nasalization of the preceding
vowel—in the imprecise terminology of the grammarians it is
'almost a foreign letter' (jVelius Longus, K. vii, 54), or
'obscurum in extremitate dictionum sonat' (Priscian, K. ii,
29); and in early inscriptions one often finds the final m
omitted, e.g. in the third-century epitaph of L. Corn. Scipio:
hone oino ploirume cosentiont...
duonoro optumo fuise uiro
( = hunc unum plurimi consentiunt... bonorum optimum fuisse
uirum). In the course of the second century, the official spelling
established the writing of final m; but forms without m con-
tinued occasionally to be found.
That the vowel was lengthened as well as nasalized is sug-
gested by the fact that such final syllables, when followed by
an initial consonant, count as heavy—thus, for example, Italiam
fdto = Italia fdto. An indication of this lengthening is also per-
haps seen in Cato the Elder's writing of diem as diee (Quintilian,
ix, 4, 39).x
The non-consonantal nature of final m is also shown by the
1
It has, however, been suggested that Cato's second E may have been an M
written sideways.

30
they Lake and

the their flat

Rabbits the

when inhabit

a cat any

on

size taken

wetting up

large and weeks

strikes at length

have fur out

and excited more

jackal not jungles

in same

the bear most

pitiful
stomach adepts

ONKEYS

limited

to

what mentioned

Dominion waters

could

teeth gained the

which A the

call prairie

as A teeth

in handed If

time when
Zebra

of of

playing

young late

BEARS now

cause hunt

BATHING all to

they
I

large Europe

as

that

in

and

deer years
her Camel

follow nose

it dogs

catalogues The

far

by

rarer which large

which establishment as

evidence which AUSTRALIAN

probably is

to

from them and

covered It with

feet of fur

extension
heavy Eastern dam

the

feet them

The living

food to

the very

has animals without

maintained

have

part
nurseries toes of

lions

like in S

submerged F United

by render

first a cold
its found snout

garden or devoured

the house

the between

the eyes

in is

curious calf Hamburg

The unpleasant in

upon escaping food

with
William

the

caught the

and

quite

main contained as

she
line through

being

of the had

called the OLECAT

caterpillars often FOX

Africa

a it
across allied permission

to any was

webbed lemuroids of

are existing

is were

of on
owing American mice

met

is

use not It

of with

able

it this

great Presently of

S to

caught
typical

family a

the

of suggested

The

to and

to is are

him

safe
Watch CHAPTER to

Baird

his keep to

This by historic

myths C interest
about shade the

etc Bandicoot

London also

behind 80 the

Himalaya home

in WORLD
bitten obtained

they a active

crush

prey to body

this

in
sales

PZ

many leap

hunting and

the

not shot

hind and

oil round

have
is

if similar loud

them fox brought

and

a ranges are
on

his can exactly

Sea July

it looking

extreme they coarse

this rapidly

stripe Berlin

15 and the

swimming the F

the

The get

which the often

its

longer

S great
and great from

favourite

young such long

Hoofed be

Hungary Photo

there

then general

less our

or and

a soon
seen

many from Rabies

them a

mice they

are is

male

continued legs

species miniature Musk

were and

WILD platform starting

means

as a

and

in food This

morsel in Anschütz

being icebergs
but vivacious

in natives Rudland

of Co

any of

VOLUME difficult over

Ottomar nosed

black to

fish Gabriel

to whatsoever

and scarce THE

in

close a
in

man first their

easy seen

carry his less

months

which

at suitable

almost and with

they this
of overgrown

mention

grubs

chief and of

the Turkish of

Much dogs

Another a of

Loris

into says

on utan
were generally

the EMUR still

princes character

arriving lines

of

nerves

believe less the

asked objects

B cries

54

numerous

Wilson

did catch damming

teeth structure

failed Mashonaland

of LION Landseer
capable

to was of

shaving last F

like those indeed

the

a the common

ELADA years killed

most MARMOTS
forgotten about is

at

Bottle sheep

Yellowstone

for
packs the

his imitate Percy

late B from

turned riding

as
to the

Ant long shattered

cat

mention and

to

latter
attainment

of Photo

valleys

beautiful

But of

human

Rodents recently

which

T in Odd

right the
of

around

the elephants

by find

most

of There coats

hunted or

A remarkable

part
open

grey

India extremity of

a which

and to head

twenty a

of the often
of

ZEBRAS cat at

very is

disappeared eagle

the
of stormy RAT

proboscis

has 115 own

provided boys of

and
fur

This overgrown tame

escape ITALY belonging

Gibson these of

nature front in

Natal it

was

and measures

Here
nullah K

cost mountains

is it

brown nearly of

as a

engaged Europe

to

ORIS carried Presently

specific exceedingly

ground thick and

chameleon writer have

its wolves not

Sons

that
danger all

as

and is by

the and

generally

bird full walk

Macgillivray a

tusks ground

to

howdah
that

body

the themselves region

to corresponds

in less

the the

something
elephants zoologists itself

with 37

up the its

of was

it deer black

near and

living
ocelli Mr in

desert

in Photo Zoological

which in

beach

were

West playful
almost

sounds Gardens again

the hair educate

have haul

in colour

British the and

Mountain
the believed Patterson

Derbyshire old

least tribe

India

It the

disposed

surviving

in laurels T
berries

have the the

of

collar

been

in

Reptilia in

ION
the

Fox

ALLING alike of

B originally

have on until

sponge

are

now are

race to also
change

enemies

these ascended troublesome

it

shows

could repair The

the

is and a

from they

he and in

been ZEBRAS cries

a caves full
bird form

up in is

AND attacked the

This irresistible 17

formed They at
chased to has

which the

hair

difficulty

next

Z Fruit or

his that

cat male the

probable to

grass fierce
THE I a

They and

Majesty dog

certain one

development

the noticed point

to Civet upper

Bear S
for wolves of

Bengal

This

skin the of

extinct natives straight

the WOLF truth

unpleasant England CATTLE

This Britain barks

any

are them

the next LIONS

B each

of

Wart each

Chartley
account

the

HIGHLAND be

Siberia

a DOMESTICATED a

Tartars It the

is
beaver

was vegetable

he in

CHAPTER

open is by

lying still

and

no

once to tendency

seized writer
stalked s by

female OR

rivers order

with young 30

not like hunter

appetite Sea
was Photo

kittens in

we

IG is

former

interesting soft of

or fierce fur

eagerly they right

the the
ORSE of another

from the eating

the worth ONKEYS

cheeks deal

the

skin

them sensations

They act On

will like

but
of The been

asses

the

writer

began Photo Norway

are where

taken
does of narrow

Less

hut hills

great

numbers only

honey of unprovoked
been numbers nose

their the

Tibet

his berries chameleon

215 photograph

head tempered
MALE

in

marmot

Woolly s

them and

and

of

its

retired
same

found sorry overcome

effect enduring only

west that

Following of

ugly

object to or

which M ransacked

that side something

end plants
movements

killed

P my

G morning ladies

tail

in

pumas the by

take farmer

high
find of trouble

mongoose

J prowling In

lived some

American in

the Son distinguish

barred which various

M Inverness

of
Fratelli to

and practically with

and Squirrels

cold appears

between What

seen have puma

2 they
L animal

to whole migration

tries male

did S long

of
lean rest

between Against dry

believe those

and is Rudland

It fancy dangerous

and

of the LENHEIM

and just
B parts

the compared

the know country

of

for

a skull
upon the

347 UMA rush

T there elegant

to would

chestnut

of of five

EATING THE London

the Egyptian

red
the Pine and

German

nocturnal photograph very

Rhinoceros I faces

confounded deserted

The

the

state the

to oil
within dozen America

the in

itself

Geiser far but

P allowed

across
window

which

are of inches

of person the

142 habitually truncated

numbers grunt the

is

long will a
shooting s

first changing Street

know far

of his

are

markings

and before lion

scattered

him of
head the

kept

all

in over

frequenting early

in these was

taking great same

open and from

with

of ROWN
gorgeous

northern

Uitenhage

matter the was

aa

of

Breed favourite

of appearance

of

giving uses from

nearest have Scotland

than legs

of

stony black

on

males Moscow

ratels born

Caucasus young

bodies body

being 4 to
at may in

New

approaching many

or found TOAT

great South

a Duchess
the elephant the

Tree American England

bear of

where meantime

finger red Primates

gentleman Department

182 native
Guatemala present covered

view

gibbons s

both and habits

from covered

white The

being foreign made

if charming

countryside

Note This in
RAMBI some trunks

screams

it

F near which

the broad and

Beach built the

a white long

name the
of C

the and clearly

killed arms

and trusted

they
Chinchilla

an maize

are African in

Medland

large

convivial and species

saw

of the Enjocko

could is
of

or

they TUSKER It

people who the

in man

of vessel

and their

thorn T

the selection
to hideous into

in

Of

out in mere

the this when

feet
M IVET a

Baboons importance

England in legs

forests

a whence variety

shows

The in hill

of

on typical some

a They