Eld Pl Csirmaz

Sailing to Byzantium
A Voyage into Symbolism

While William Butler Yeatss Sailing to Byzantium is often described as less complex than Byzantium, the differences between the two poems appear to have
rarely been considered on levels other than meaning or referents. This essay aims to
unearth a basic textual difference which may account for the above judgement with
the help of a framework rooted in structuralism and inuenced by the theories of
Alexander A. Potebnja. The analyses of the two poems allow the conclusion that
while Byzantium can be regarded as a symbolic text, Sailing to Byzantium approximates that mode of writing without being entirely controlled by it.

Introduction
To one scanning a selection of the overwhelming amount of scholarly work
prompted specically by the two Byzantium poems of William Butler Yeats, Sailing
to Byzantium and Byzantium, it might appear that, faced with two texts of multiple interpretations, analyses often turn to sources outside the poems in order to
underpin a reading or a paraphrase. Such sources might include biographemes from
Yeatss life, other writings of Yeats (in an attempt to regard Yeatss quasi-mythology
as a largely unied system), or historical accounts of Byzantium and Ireland. T. R.
Henn, in his 1965 book, went as far as suggesting that Byzantium deals with the
poets real or imagined loss of sexual power.1 In like manner, differences between
the two poems, if they are addressed at all, are treated as differences in meaning,
reference or degree.
It is on the level of theme that Richard J. Finneran captures the main dissimilarity between the texts. According to him, Sailing to Byzantium explicates the
reasons for the necessity of Art, while Byzantium considers how a work of art is
1. T. R. Henn, Byzantium, in The Lonely Tower: Studies in the Poetry of W. B. Yeats
(London: Methuen, 1965), 220237, p. 236. See also Brenda S. Webster, A Psychoanalytic
Study: Sailing to Byzantium, in Yeats: A Psychoanalytic Study (Palo Alto, CA.: Stanford
University Press, 1973); reprinted in Literary Theories in Praxis, ed. Shirley F. Staton (Philadelphia: University of Pennsylvania Press, 1987), 307312.
The AnaChronisT 14 (2009): 88110 ISSN 12192589

SAILING TO BYZANTIUM

created.2 Marjorie Perloff, who bases her analyses on formal considerations (rhyme
pattern), also places the difference on the level of meaning, suggesting that Sailing
to Byzantium expresses tensions, as opposed to Byzantium, which is characterized by reconciliation.3 F. A. C. Wilson, capturing the difference on the level of symbols, sees the central element, Byzantium, as referring to the realm of intellect and
spirit in Sailing to Byzantium, while the same word, in Byzantium, symbolizes
life after death.4 Just as the nature of the underlying troubling experience and the
interpretation of symbols in these two poems remain unclear, so it remains uncertain to what Byzantium may actually refer in the texts. Finneran enters into polemics with Frederick L. Gwynn, who argues that the city in Sailing to Byzantium
is taken from its real-world counterpart in the early sixth century (which, as he
suggests, might correspond to Yeatss Phase 15), while the Byzantium in Byzantium is modelled on the city as it stood around 1000 (which may correspond to
Phase 28).5 By contrast, Finneran, considering other sources, suggests that in both
poems the model of the city was Byzantium in the 900s.6
According to a cursory glance at a portion of the analyses of the Byzantium poems, such are the differences which appear to have been considered most often
apart from passing remarks on the general nature of texts. These remarks treat
Byzantium solely as a sequel to, or even a development over, the previous poem,
Sailing to Byzantium. G. S. Fraser remarks that Sailing to Byzantium is rather
abstract compared to Byzantium;7 and while T. R. Henn adds, in parentheses,
that Byzantium wears less well than the other [poem], he also suggests that in
Byzantium the system of tensions is more complex, the overtones more
signicant.8 John Unterecker simply states of Sailing to Byzantium that it is
clear enough,9 while A. G. Stock claims that it explains itself.10 Finneran, it
2. Richard J. Finneran, introduction to William Butler Yeats: The Byzantium Poems, ed.
Richard J. Finneran, The Merrill Literary Casebook Series (Columbus: Charles E. Merrill
Publishing Company, 1970), 110, p. 6.
3. Marjorie Perloff, The Rhyme Structure of the Byzantium Poems, in Rhyme and Meaning in the Poetry of Yeats (New York: Humanities Press; The Hague and Paris: Mouton,
1970), 122131, 141143; reprinted in Literary Theories in Praxis, 2132, p. 21.
4. F. A. C. Wilson, W. B. Yeats and Tradition (London: Victor Gollanz, 1958), p. 231.
5. Frederick L. Gwynn, Yeatss Byzantium and its Sources, Philological Quarterly XXXII
(January 1953) 921, pp. 1011.
6. Finneran, pp. 35.
7. G. S. Fraser, Yeatss Byzantium, Critical Quarterly II (Autumn 1960) 253261, p. 256.
8. Henn, pp. 236, 228.
9. John Unterecker, A Readers Guide to W. B. Yeats (New York: Noonday Press, 1959), p.
171.
89

ELD PL CSIRMAZ

seems, agrees to a certain degree, admitting himself that it is clearly a simpler

poem than its companion-piece.11 Anthony L. Johnson also deems Byzantium to
be more complicated, but not at all to its advantage: he suggests of Byzantium that
its obdurate resistance to glossing has been erected by some critics into a patent of
aesthetic merit.12
I do not think, however, that such remarks (often bordering on value judgments) are to be done away with. Instead, they should be treated as genuine responses from educated readers, and textual evidence should be searched for to
account for their apparent congruence.
In this essay, I will analyse both Byzantium poems. The aim of my analyses is
to point out an underlying, basic difference between the two texts, which may also
account for why Sailing to Byzantium strikes most readers as simpler or less complex than Byzantium. This difference, however, shall not be sought on the levels of
interpretation or referentiality, but on the textual level, or, one might say, in the
structural set-up of the poems. I will argue that while Sailing to Byzantium does
converge toward symbolism, Byzantium is a genuinely symbolic poem, especially
if read against the previous text. I am not, however, suggesting that Byzantium is
a development over the previous poem, especially as the differences between the
two texts point in various directions.
In my analyses I will primarily adopt an approach based on the reception of the
poems by a present-day reader. I will, in other words, limit references to the biographical Author, either his life, other writings, or knowledge, or a (re)constructed
compositional process.13 Instead, I attempt to focus on the texts themselves (their
nal versions), and seek intra- and inter-textual features, similarities and dissimilarities. The relations to be pointed out may, I hope, further our understanding of
the workings of the two poems and their continuingly strong effect on us as readers.
The concepts and strategies I will utilize in my readings are derived partly from
Alexander A. Potebnjas theory of literature and partly from other structuralist approaches. For easier reference, I have coined the phrase representational framework
10. A. G. Stock, W. B. Yeats: His Poetry and Thought (Cambridge: Cambridge University
Press, 1961), p. 202.
11. Finneran, p. 7.
12. Anthony L. Johnson, Sign, Structure, and Self-Reference in W. B. Yeatss Sailing to
Byzantium, Annali della Scuola normale superiore di Pisa, Classe di lettere e losoa 8
(1978) 213247; reprinted in Literary Theories in Praxis, 135154, p. 135.
13. Yeatss manuscripts have been considered in A. Norman Jeffares, The Byzantine Poems of W. B. Yeats, The Review of English Studies 22.85 (January 1946) 4452; a meticulous analysis has been provided by Curtis Bradford, Yeatss Byzantium Poems: A Study of
Their Development, PMLA 75.1 (March 1960) 110125.
90

SAILING TO BYZANTIUM

to refer to this approach. Scrutinizing it further in application has, in fact, been an

indirect aim of the following analyses. The framework focuses both on larger structures or images and on their building blocks, on elements on the level of words. This
latter facet, which echoes close reading, is anticipated by David I. Masson, who, in the
rst half of his treatise, considered repeated words and their contexts in Byzantium
in a way reminiscent of Claude Lvi-Strausss analyses of myths based on mythemes.14
Johnsons structuralist analysis is also instructive in this respect.
Of image-based reading, precisely what I set out to provide of the two poems,
William Empson declares: I nd here a breath-taking assumption that language
can never be used to tell a story; that the only literal meaning it can have is the
pictures it makes in your head.15 Empsons intention was to criticize F. A. C. Wilsons approach to Yeatss poetry in W. B. Yeats and Tradition, claiming that Wilson
disregarded literal meaning in favour of a symbolic one. And yet I think that albeit
unintentionally Empson here gave a key to an understanding of the workings of
the Byzantium poems. I agree with Caroline Roberts, who sees Yeatss language as
one that demonstrates that all language is irreducibly gural.16 Let me start my
analyses with a brief introduction of the framework I will attempt to utilize.

The Representational Framework17

This framework, generally speaking, postulates that the extra-textual author gains
representation in the form of an intra-textual author or the lyrical I, and that the
original experience of the extra-textual author gains representation in the text in a
visual model centred around subjects, or, as I shall refer to them, exhibits. Such a segregation of extra-textual and intra-textual layers along with certain categories I shall
introduce may strike one as outmoded from a post-structuralist point of view. However, these concepts and categories prove to be exible enough to incorporate suggestions that have been used to criticize structuralist approaches to literature.
Parts of the framework have been inuenced by the theory of Alexander A. Potebnja on the threefold structure of literary works, which has been presented and
14. David I. Masson, Word and Sound in Yeats Byzantium, ELH 20.2 (June 1953) 136
160.
15. William Empson, Mr. Wilson on the Byzantium Poems, A Review of English Literature I (July 1960) 5156, pp. 5354.
16. Caroline Roberts, Images Fresh Images Beget: Yeats, De Man, and Deconstruction,
in Yeats: An Annual of Critical and Textual Studies XVI, ed. Richard J. Finneran (Ann Arbor, MI.: University of Michigan Press, 2001), 8499, p. 85.
17. The following concepts were inuenced by va Babitss views and theories presented
during discussions between 1997 and 2001.
91

ELD PL CSIRMAZ

meticulously analysed in a monograph by John Fizer.18 This framework postulates

the existence of three layers both in the process of composition and in that of reception of a work of art; the processes are regarded symmetrical.
Layer III is, on the rst level of approximation, situated outside the work. In
the compositional process, it is the original experience, problem, etc. experienced
by the author which becomes represented according to this model in the work
of art. In reception, layer III is the interpretation created by a reader. Generally, it
is supposed that the original experience of the biographical author is irrecoverable from the text and that interpretation differs among readers. Layer I is the
textual level. It is the message transmitted during the artistic communication, but
it can also be considered to contain connotations and associative elements generally available to members of the interpretive community in which the artistic
discourse is taking place. And, nally, layer II is dened as the structures generated by the text in the reader during the act of reading. Theoretically, this layer is
supposed to contain all possible structures; I suggest that it is during interpretation (generation of layer III) that readers emphasize certain structures in line
with their interpretation, and suppress others. This suggestion could be complemented by the idea that the generation of layer III might be optional at least in
the sense of formulating a conscious and verbalized (thus necessarily abstracted)
interpretation of a work of art. Layer II, therefore, has a dual nature. On the one
hand, strictly speaking, it exists only in the reader and is a result of reading; on
the other, during its generation, in theory, no inter-reader differences have yet
been introduced. It is supposed, in other words, that layer II is identical across
readers and that it contains the same structures in both the reception and composition processes.
In Potebnjas theory, a similar set of three layers can be found: the external
form (cf. layer I), the internal form (layer II) and signication, content or idea (layer
III).19 His views agree with mine in suggesting that signication changes markedly
in every new perception;20 in that the internal form tends to expire, reducing the
structure of literary works to two constituents external form and signication
and its potential polysemy to a referential monosemy; and in that one of the main
18. John Fizer, Alexander A. Potebnjas Psycholinguistic Theory of Literature: A
Metacritical Inquiry (Cambridge, Mass.: Harvard Ukrainian Research InstituteHarvard
University Press, 1988).
19. Fizer, pp. 23, 37.
20. Alexander A. Potebnja, [Notes on the Theory of
Verbality] (Kharkiv: Izd. M. F. Potebni, 1905), Slavic Printings and Reprints, 128 (The
Hague: Mouton, 1970), p. 101; quoted in Fizer, p. 23.
92

SAILING TO BYZANTIUM

differences between poetic and non-poetic (scientic, in the extreme) texts is that
the latter lack internal form.21
Despite these similarities, there are also a number of differences between my
framework and Potebnjas theory. Most importantly, Potebnja appears to have embedded the above-cited ideas in an inherently Romantic and 19th-century view of
language, suggesting, for example, that language mirrors and originates from perception.22 Moreover, in his view, the poetic work contains its internal form, which,
like the external one, is internalized during the act of reading.23 In opposition to this
view, I hold that it is only the external form (layer I) that is transmitted and internalized in the strict sense. The linguistically coded nature of layer II in Potebnjas
theory possibly makes more sense if one considers his suggestion that the word has
a threefold structure identical to that of literary texts24 a notion entirely missing
from the framework I am presenting.
Similarly to Potebnja, I regard as one of the central structural building blocks of
layer II the image. In the present framework, it functions as a sample structure with
which an attempt is made to analyse and describe structures in layer II. For Potebnja, however,
while it was relatively simple to dene the internal form of the word, inasmuch as Potebnja equated it with its etymon, the image of the work of poetic art eluded an easy denition. His theory, in spite of the central
importance of internal form, gave no denition of the image.25
My denition of the image is relatively simple and exible. It is a set of related
textual elements which describe that is, create a model which can be visualized
the element at the centre of the image, the exhibit. The image and its exhibit are said
to be motivated by elements related to them. The elements that I consider capable
of motivating images are those which are able to trigger immediate (usually visual)
associations. I shall use concreteness in the sense that a concrete element triggers
more associations than an abstract or conceptual one. These properties do not imply a difference of character; rather, they represent a one-dimensional graded cate21. Fizer, pp. 27 (see also 33), 36.
22. These ideas pervade his writings: [Thought and Language],
[Notes on the Theory of Verbality], and
[Lectures on the Theory of Verbality], trans. to Hungarian G. S.
Horvth, K. Szitr, K. Horvth, A. Molnr, K. H. Vgh, in Potika s nyelvelmlet [Poetics
and Language Theory], ed. rpd Kovcs (Budapest: Argumentum, 2002), 5250.
23. Fizer, p. 47.
24. Fizer, passim, see especially p. 37.
25. Fizer, pp. 4041.
93

ELD PL CSIRMAZ

gorization. Concreteness, moreover, is essentially the same as the level of abstraction of names of objects (categories) as dened in cognitive science, and as used,
among others, by Eleanor Rosch.26 According to her denitions, when two object
names (categories) are related to each other via class inclusion, it is easy to determine their relative concreteness, as greater inclusiveness means a higher level of
abstraction. To take an example, swallow is more concrete and less abstract than
bird, as the latter contains the category swallow, and the subordinate category is
not exhaustive of bird, the super-ordinate one. This implies that swallow has a
more specic size, outline and plumage than a generalized bird, even if prototypicality is taken into account. Swallow is also associated with summer, based on
the idiom, which contributes further to the associations this element can trigger. In
a similar fashion, short noun phrases (e.g. kitchen table, stone table) are more concrete than what they specify (table).
When, however, none of the elements (categories) are included in the other, an
indirect comparison has to be made. Rosch introduces the notion of basic objects
(e.g. table) which are the most inclusive categories whose members still share a
large number of attributes. She also suggests that most basic objects are at the same
level of abstraction.27 I consider basic objects the most abstract elements that can
motivate images, as super-ordinate categories (e.g. furniture), that is, concepts, are
usually not sufciently able to be visualized. In fact, Rosch suggests that basic objects are the most inclusive categories which can have a mental image.28 With basic
objects dened, it is possible to compare the concreteness of two unrelated elements
by comparing their perceived distance from, that is, their inclusiveness relative to,
their respective basic objects.
Regarding grammatical categories, I have found that elements motivating images are most often nouns or short noun phrases. However, certain verbs (consider
ghting or plucking, for example) or other content words may also be visual enough
to be included in this class.29
26. Eleanor Rosch, Principles of Categorization (1978), in Concepts: Core Readings, ed.
Eric Margolis and Stephen Lawrence (Cambridge, MA: MIT Press, 1999), 189206.
27. Rosch, pp. 192, 191.
28. Rosch, p. 195.
29. Rosch herself extended the notion of basic objects to events. Moreover, concreteness
and abstractness may also be related to the amount of information a given element conveys.
In this respect, swallow is more concrete than bird because it selects a more limited set of
entities. In order to restrict the number of entities an element refers to, it needs to provide
more information regarding the entities themselves, and a greater amount of information
enriches an image to a greater degree.
94

SAILING TO BYZANTIUM

Analysing the distribution of these elements has proved fruitful partly because
they can be regarded as unaltered during their translation from layer I to layer II in
the perception process. In this respect, these elements behave like minimal, atomic
building blocks, and based on this property, I also refer to them as minemes. It can
also be suggested that the set of elements associated with a mineme is more or less
xed in a given interpretive community, while, at the same time, such associations
will, naturally, be altered by the context, by other minemes in the same image.
This framework goes further, however, in forcing a prescribed structural set-up
on layer II. It postulates that images themselves are ordered hierarchically, allowing
images to be part of, or to motivate, larger images than minemes do. Images at the
centres of these hierarchies are basic images. Sometimes it is possible to select one
basic image for a whole work, the global basic image, the exhibit of which is the
global exhibit of the text.30
It is also worth noting that while Potebnja himself did not dene the image,
Fizer attempted to abstract a denition from his arguments. According to him, Potebnja regarded the construction of images as happening either step by step, combining representations in words, or suddenly, at certain points in the text, where the
internal form of a word dominates those around it.31 My denition of the image appears to be a combination of these two modes inasmuch as every image is postulated
to have a centre, while it is enriched by a series of other elements at the same time.
I dene the theme of a work of art as its experience or interpretation abstracted
to a level which is common to all interpretations and the experience. (In this
framework it is supposed that it is possible to do so, based on the similarity of readers in an interpretive community.) Metathesis refers to the relationship between the
represented theme / experience / interpretation in layer III and the representing
layers II and I. In other words, it relates the experience centred around the theme to
the representation centred around the global exhibit of a work. If there is no metathesis, that is, if layer III is rendered directly into layer I, and layer II is missing,
then the text is considered to be non-artistic.
The lyrical I is the intra-textual addresser of the message in the artistic communication. I termed the relationship between the author and its representation
the lyrical I alienation mainly because I generally suppose them to be distinct
and connected by nothing else than the representational relationship. I regard metathesis and alienation as parallel and hardly separable processes, as one describes
rendering the object, the other the subject of an experience into the object and the
30. Despite certain differences, Potebnjas main image might be related to what I mean
by basic images. See Fizer, p. 45.
31. Fizer, p. 48.
95

ELD PL CSIRMAZ

subject of a representation. Thus, as far as the scope of the present framework

reaches, alienation, similarly to metathesis, is required for artistic texts. This requirement echoes the notion of the death of the author as a necessary step (technique) in writing. T. S. Eliot and Mark Schorer, among others, refer to the alienation
of the experience from the original experiencer, the author, as essential in artistic
creation.32
Images, however, are not the only tools with which the distribution of minemes
in layer II can be investigated. Elements can be divided into two categories according to the roles they play in gures as similes, metaphors and symbols.
The descriptive relationship between two elements in a simile or a metaphor is
asymmetrical. While both the described and the describer elements (or the tenor
and the vehicle in the case of metaphors) are evoked, it is the implicit goal of these
gures to provide further information about the described using the describer.
While it does happen that the describers build up a hidden textual layer inside the
work, or that the described and the describer are reversed, these techniques are
notable especially because they deviate from the norm. The describer is thus usually
secondary to the described and is referred to only in the context of the latter, while
it is the described that the focus falls on and is understood as the one that needs to
be placed alongside other elements in the text. Take, as an example, the following
sentence: He winced at the sight of the laundry as his hand was already as dry as the
back of a camel after a weeks journey in the Sahara. In this very basic case of an explicit simile, it is immediately understood that the focus falls on the hand, and that
while a vivid image of the desert might be evoked, it is not to be aligned with laundry,
wincing and hand. In other words, all minemes in the describer are removed from
the sphere of the described, and they are hardly ever thought of as being in a physical connection with them. Referring to the imaginable physical environment of an
element as its reality, this means that the describer is generally absent from the
reality of the described, and that the described is almost always considered to belong to the reality of its larger textual context or to that of the text itself.
Taking this statement to the extreme, by investigating various gures in a text it
may be possible to determine which elements create its reality, and which are absent from it; in other words, which minemes function more as describeds and which
more as describers. Following va Babitss terminology, I will refer to the class of
describeds as view or layer A, and to the class of describers as vision or layer B.
32. See T. S. Eliot Tradition and the Individual Talent (1919), in 20th Century Literary
Criticism: A Reader, ed. D. Lodge (London: Longman, 1972), 7177; and Mark Schorer,
Technique as Discovery, in The World We Imagine (New York: Farrar, Straus and Giroux,
1968), 323.
96

SAILING TO BYZANTIUM

As I attempted to show in my paper titled Incorporation and Dissociation:

Changes in the A/B Structure between Realism and Modernism,33 the gure of the
symbol can be analysed as a describer element originally belonging to layer B but
moving to layer A to take the place of an absent described. This proposed analysis is
not without parallels. It is, among others, closely related to Roman Ingardens remarks on symbolization:
It is part of the essence of the symbolizing function that (1) what is symbolized and that which symbolizes it belong to different worlds . . . (2) what is
symbolized is in fact only symbolized and cannot attain self-presentation.
As something symbolized, it is, according to its essence directly inaccessible, it is that which does not show itself.34
In the same paper, I also argued that myths essentially have a similar internal
structure, which suggestion, in turn, can be related to the observations on the nature of myths by S. A. Tokarev and Y. M. Meletinsky, who argue that symbolism is
the most important feature of myths, which manifests itself in the unclear distinction between subject and object, between object and sign. They also state that concrete objects, without losing their tangibility, may become signs of other objects or
phenomena, that is, they may replace or represent them symbolically.35 These two
propositions will, in fact, make it possible to assess the symbolic nature of the texts
to be scrutinized.

Sailing to Byzantium
Sailing to Byzantium was written in 19261927 and rst appeared in October
Blast in 1927 and later as the opening poem in Yeatss volume of poetry titled The
Tower in 1928. As each stanza of the poem can be read as a self-contained image or
pair of images, let me start studying the text by considering its stanzas one by one.
The minemes found in the second sentence of the rst stanza which spans
almost the whole portion motivate an image describing a series of habitats bursting with life (see, for example, young, birds, salmon, mackerel, summer) and
temporality, as added by the more conceptual phrases in lines 3 (Those dying gen33. In Els Szzad 2006/2: 81102.
34. Roman Ingarden, The Literary Work of Art: An Investigation on the Borderlines of
Ontology, Logic, and Theory of Literature, trans. George G. Grabowicz (Evanston, Ill.:
Northwestern Univ. Press, 1973), pp. 299300.
35. These notions are taken from S. A. Tokarev and Y. M. Meletinsky, Mitolgia, in Mitolgiai enciklopdia, ed. S. A. Tokarev et al., trans. Gyrgy Brny et al. (Budapest: Gondolat, 1988), 1:1121, p. 13.
97

ELD PL CSIRMAZ

erations) and 6 (Whatever is begotten, born, and dies).36 With this image is contrasted the mineme old men, which appears to be aligned with the monuments of
unaging intellect (8). These, in turn, are described as being in opposition to the
life-image built up during the course of the stanza. This dualistic arrangement of
elements will be perceivable throughout the poem, with nearly all minemes clearly
belonging to the life or the monument (art) side of the world-model the poem appears to suggest.
The exhibit of this stanza may be young, as it appears in the rst line, as it is
this mineme that in itself carries the notion of life and reproduction which all other,
connected elements point to; and which the mineme old men is contrasted with.
The use of sh, esh, or fowl in line 5 is worth a closer perusal, especially in the
light of this suggestion. It is not only a list of elements easily merging with birds,
salmon or mackerel, but which also span a whole in the sense of all of a type of
thing, as the negative of this list, it is neither sh, esh, nor fowl means a food t
for no class of people.37
The second stanza does not offer itself easily to an image-based reading. There
are few words or phrases that are undoubtedly analyzable as minemes (aged man,
tattered coat, stick, hands, tatter, dress, monument, to list the ones which, in my
view, are more capable of visualization), and most of these elements describe the
aged man employing a metaphor that spans the rst two lines (An aged man is but
a paltry thing, / A tattered coat upon a stick), and which is also enriched by the
introduction of the soul and its mortal dress. This latter element, through the repetition of the notion of tatter (1112), is linked to the element the metaphor described, the aged man. This reading implies that the element most motivated by this
scarecrow-image is the aged man, which, thus, serves as the exhibit of this portion
of the poem. The second half of the stanza, however, contains almost no elements
easily capable of visualization. The fact that it is at this point that the rst reference
to the lyrical I (And therefore I have sailed the seas [15]) is found will be of special importance when analysing the overall structure of the poem.
The dualism of the rst stanza continues in this one as well. The set of minemes
consisting of a song-less aged man, tattered coat, stick and mortal dress is contrasted with singing, for An aged man is but a paltry thing . . . unless / Soul clap its
hands and sing (912). Singing, in turn, is linked to the monuments already introduced at the end of the previous section: Nor is there singing school but studying /
36. References to the poems are to W. B. Yeats, Sailing to Byzantium and Byzantium, in
Yeatss Poems, ed. A. Norman Jeffares (London: MacmillanPapermac, 1989), 301302,
363364, respectively. All parenthesized references are by line numbers.
37. E. Cobham Brewer, Dictionary of Phrase and Fable (Philadelphia: Henry Altemus,
1898; Bartleby.com, 2000), s.v. sh3 <http://www.bartleby.com/81/6492.html>.
98

SAILING TO BYZANTIUM

Monuments of its own magnicence (1415). With the ending, it becomes apparent
that such monuments are to be found in Byzantium.
The dualistic distribution of minemes also characterizes the third stanza, where
the sages, aligned with gold, mosaic, wall, re, gyre, perne38 and the concepts
artice and eternity are set in opposition to the heart of the lyrical I, linked to
mortality, to a dying animal (2122). The elements motivating an image of an eternal artwork, the mosaic, are introduced by a rhetoric gure, a simile, as the sages
[are] standing in Gods holy re / As in the gold mosaic of a wall (1718, my emphasis). Because of this fact, and in line with the denition of layers A and B, the
mosaic and the elements related to it will reside in layer B.
One nds the same set-up in the last stanza, where the image of the golden
handiwork is introduced as a form as Grecian goldsmiths make (27, my emphasis)
with the help of a simile continuing for three lines. This image is enriched further by
a reference in the last three lines of the poem probably to an automaton of Byzantine Emperor Theophilos Ikonomachos. This automaton consisted of articial birds
on a tree, emitting various sounds, all of gilded bronze;39 which are thus of hammered gold and are set upon a golden bough to sing (28, 30).
Determining the exhibit of the previous stanza, the third might be problematic,
as this section is the one which yields least to a reading based on minemes. Once
38. Gyre and perne may only be partially decipherable for a reader less familiar with
Yeatss personal mythology and terminology, and thus, strictly speaking, any reading beyond
the literal meaning of gyre should be excluded from an analysis which attempts to disregard
information related to the biographical author. This suggestion is supported by the fact that
the editors of The Norton Anthology of English Literature felt it necessary to explicate the
phrase in question in a footnote, attributing to it the meaning to whirl round in a spiral motion based on a reading of perne (pirn) as bobbin. This motion is then connected to the allencompassing force of fate and history in Yeatss system (The Norton Anthology of English
Literature, 7th ed., ed. M. H. Abrams et al. [New York, London: W. W. Norton, 2000], 2:2110).
39. Liudprand reported on this automaton when sent on a mission to Constantinople in 949.
See Liutprandus Cremonensis, Antapodosis, in Patrologia Latina, ed. Jacques Paul Migne
(Paris, 1854), vol. 136, column 0895. I thank Anna Tsks for her kind help in locating this
source. Additionally, a long list of possible sources has been compiled for the articial bird imagery in the two Byzantium poems based on various considerations. Jeffares, who was advised
by Mrs. Yeats on her husbands readings, ultimately decides that the exact source cannot be
identied. Thomas L. Dume, in his Yeats Golden Tree and Birds in the Byzantium Poems
(Modern Language Notes 67.6 [June 1952] 404407), lists a number of books Yeats probably
read; the list is lengthened by Gwynns account of possible analogues in other works (see especially pp. 1318). Perhaps one of the most intriguing suggestions is made by Ernest Schanzer,
who, in an essay titled Sailing to Byzantium, Keats, and Andersen (English Studies XLI [December 1960] 376380), links the image to Andersens tale The Emperors Nightingale.
99

ELD PL CSIRMAZ

again, this characteristic coincides with a prominent textual presence of the lyrical
I. Still, as most minemes motivate the image centred around the sages (consider
re, gold, mosaic, wall), this may be the exhibit searched for.
Within stanza four, determining the exhibit is an easier task, as the text culminates in the description of the automaton, of which the most important part is what
the lyrical I strives to become, an articial, gilded bronze bird. As this concrete image
ows naturally from and species the previous, more general form as Grecian goldsmiths make (27), I regard bird as the exhibit. It has an inverse relation with the lyrical I, as with the desired transition to the eternal state direct references to the lyrical
I gradually disappear from the text, as if it were dissolved in the anonymity of eternity. The last reference, a personal pronoun, is to my bodily form (26).
The temporal dimension of the poem, as shown by the exhibits of the stanzas,
spans almost a human lifetime. From the young of the rst stanza, the description
moves to the aged man, culminating at the sages of stanza three, who are already
situated half in the natural, half in the eternal world. By the beginning of the fourth,
the status of being out of nature is already reached by ironically and in a circular
manner the exhibit bird, an originally natural element.
As an aside, let me point out that the life cycle as presented in this poem might
be related to a whole cycle in Yeatss system of Lunar Phases ending in complete
objectivity, in, perhaps, the sages, monuments and Byzantium. This would imply
that the poem is not passively stuck in Byzantium as Phase 15, as Frederick L.
Gwynn suggests, but provides a dynamic model of this facet of Yeatss system.
From the temporal perspective, the appearance of the lyrical I in the second
stanza has a particular signicance, as it denes the lyrical I as similar to the aged
man the stanza in question describes. This placement of the lyrical I is further
supported by the slightly distanced depiction of the country of the young with which
the poem begins. Moreover, in earlier drafts of the text, this placement was even
more apparent. In the composition process of Sailing to Byzantium, as reconstructed by Curtis Bradford, the successive drafts show not only an eradication of
specically Irish allusions and the abandonment of the poet of the Middle Ages
that Bradford sees as a model for the lyrical I, but also a gradual obliteration of
direct references to the lyrical I. A rst version of a portion that made it to the
published poem contains such a large number of such direct references as to make it
clear that the poem evolved from a text with a highly personal tone:
As in Gods love will refuse my prayer
When prostrate on the marble step I fall
And cry aloud I sicken with desire
And fastened to a dying animal
100

SAILING TO BYZANTIUM

Cannot endure my life O gather me

Into the artice of eternity.40
Determining the global exhibit of Sailing to Byzantium is hindered by the fact
that the stanzas present themselves as self-contained, separate images; and almost
none of their exhibits is motivated by the other images which could elevate a local
exhibit to the global level. There are, however, two related elements which reappear,
in one form or another, in most images: bird and song. While strictly speaking bird
appears only in the rst and in the last stanzas via the reference to the Byzantine
machinery, the element of song, if the verbal form is regarded as an equivalent, does
appear in each image. Based on this fact, I regard the birdsong complex as the
global exhibit of the poem, the elements placed at the centre of layer II. Furthermore, these are the two elements that do not conform to the dual arrangement of
minemes. In the rst stanza, both bird and song are aligned with the ageing, natural
world (see birds in the trees / Those dying generations at their song [23]),
while as early as in the second stanza singing is presented as something that lifts
the mortal old man out of its earthly existence: An aged man is but a paltry thing
. . . unless / Soul clap its hands and sing (911, my emphasis). In the fourth stanza,
the metal bird is precisely what the lyrical I strives to become in order to be preserved outside nature. In other words, the birdsong complex, the global exhibit,
bridges the two realms, the mortal, natural world and the eternal world of art.
Although analyses do not agree on the meanings and referents of specic elements, and on the existence and nature of a more general idea behind contrasting
art with nature and mortality - that is, the ephemeral with the eternal - they all
seem to agree that this dichotomy is a central theme in the poem, as shown also by
the dual arrangement of minemes. In some respect, the same underlying theme
appears to be elaborated in John Keatss Ode on a Grecian Urn, but, as William
Empson points out, Yeatss poem may contain a pinch of self-mockery as the lyrical
I strives to become a clockwork dickey-bird . . . scolding like Donald Duck.41
Richard J. Finneran interprets the last four lines in a similar way, suggesting
that to sing to keep a drowsy Emperor awake and to lords and ladies of Byzantium (29, 31) demeans the articial bird and its desired eternal state in Art. Moreover, it would sing of mutability, of what is past, or passing, or to come (32), hence
we are led back to ephemeral nature in a complete cycle.42
Incidentally, this reinterpretation of bird more as a referential, even natural
thing as opposed to a kind of an emblem - which state it reached in the Byzantine
40. Bradford, p. 113. I have omitted phrases deleted in the manuscript.
41. Empson, p. 55.
42. Finneran, pp. 67.
101

ELD PL CSIRMAZ

setting - may provide a miniature model of Paul de Mans view on the progress of
Yeatss oeuvre as rst moving from natural imagery to symbols/emblems; but then
Yeats soon wearies of a purely emblematic style, dismisses it as allegory or mere
embroideries and returns, after 1900, to what seems to be a more natural kind of
image.43
In summary, the poem does not appear to be image-driven in the sense that it
contains passages which resist a mineme and image-based reading, and the minemes themselves appear to be more abstract. It will be of special importance that,
according to the separation of minemes between layers A and B, the description of
Byzantium resides in its entirety in layer B.

Sailing to Byzantium
The cyclical structure of Sailing to Byzantium outlined above may be behind the
most often cited reactions to the poem. Yeats appears to have received this criticism
on Sailing to Byzantium from his friend T. Sturge Moore, which, ultimately, appears to have led to the writing of Byzantium. Moore wrote: [Sailing to Byzantium] lets me down in the fourth [stanza], as such a goldsmiths bird is as much
nature as a mans body.44 Indeed, the element bird, as it bridges the natural and
the eternal realms, is partly in nature. Moores criticism, therefore contrary to a
widespread suggestion might not have been seen by Yeats as an observation that
invalidated the structure of Sailing to Byzantium. It might have merely meant, as
Yeats himself wrote to Moore, that the idea needed exposition.45 Finneran, along
similar lines, suggests that Moore had explicated the main idea of Sailing to
Byzantium rather precisely while thinking that the poem was attempting to say
something else.46
Byzantium, written in 1930, does reiterate many of the elements of Sailing to
Byzantium; moreover, its internal structure shares some characteristics with the
other poem. In Byzantium, the stanzas also seem to motivate separate images, but
they are connected by more than a temporal link and the recurrence of certain minemes connections that could also be found in the earlier poem.
At the beginning of the rst stanza, we nd the element of the Emperor, whose
soldiery are abed (2). The description of dusk already contains a dual set-up of
43. Paul de Man, Image and Emblem in Yeats, in The Rhetoric of Romanticism (New
York, Columbia University Press, 1984), 145238, p. 172.
44. Finneran, p. 5, quoting W. B. Yeats and T. Sturge Moore: Their Correspondence,
19011937, ed. Ursula Bridge (New York, 1953), p. 162.
45. Finneran, p. 5, quoting W. B. Yeats and T. Sturge Moore, p. 164.
46. Finneran, p. 5.
102

SAILING TO BYZANTIUM

elements reminiscent of the two contrary worlds of Sailing to Byzantium. On the

one hand is a starlit or moonlit dome (5) which disdains things placed on the
other: man, complexities, fury, mire, veins. These latter elements motivate a subimage of mortal life in the same manner as the rst stanza of Sailing to Byzantium
contains a description of life in nature. Opposed to this image one nds celestial
elements (star, moon) and an articial, artistic element serving as the exhibit of this
portion of the text (dome), which may evoke the idea of eternity as represented
in/by undecaying, seemingly timeless phenomena.
The second stanza starts with a rhetorical device that will be repeated in the
third. The rst two lines, Before me oats an image, man or shade, / Shade more
than man, more image than a shade (910), introduce the description of an image
centred around the element of the mummy, which includes its spiral wrappings:
Hades bobbin is bound in mummy-cloth (11), and it may unwind the winding
path (13). The result is an image-complex that reiterates the notion of spiralling
movement from Sailing to Byzantium (c.f. perne in a gyre [19]), and therefore
turning to the extra-textual author may be related to Yeatss idea of the gyre,
which controls fate and the progress of the world. The presence of bobbin like that
of perne (possibly pirn, bobbin) in Sailing to Byzantium may reinforce this connection by evoking the Fates, who control the thread of life.
It is this mummy-image that is further motivated by the introduction of the
element mouth, itself serving as a centre to a sub-image spanning lines 1314. This
image, which may refer to the Opening of the Mouth ritual in ancient Egypt, is
that of a mouth which is devoid of its crucial characteristics, as it has no moisture
and no breath like a mouth of a mummy , and which, as it may summon
breathless mouths (14), appears to have been conjured up by a member of its own
kind - which, in other words, is described as autogenetic.
It is in this stanza that the lyrical I appears for the rst and last times in the
poem. Line 9 contains a reference to me, before whom the mummy-image appears, and lines 15 and 16 make it clear that the lyrical I approves of the appearance of the vision: I hail the superhuman. With the verbs hail and call, these lines
(along with the beginning of the stanza) almost merge the lyrical I with the implied author during the act of composition, for it is the author before whom the
vision appears that is transformed into the text.
The third stanza starts with the same rhetorical device as the second, rst listing elements joined by the conjunction or, then, in the second line, expressing a
preference for one element over the others: Miracle, bird or golden handiwork, /
More miracle than bird or handiwork (1718). In this case are introduced the elements bird and gold, which have also played important roles in the previous poem.
Bird here is described in more or less the same manner as in the last stanza of Sail103

ELD PL CSIRMAZ

ing to Byzantium, with the same reference to the Byzantine construction. What is
notable here is the recurrence of the elements star and moon, as the articial bird is
planted on the star-lit golden bough (19) and is by the moon embittered (21). In
the rst stanza, these very elements were used to describe the dome, which was
contrasted with man described by complexities, fury, mire and veins here repeated as complexities, mire and blood in lines 2324, now related to the common
bird. In other words, the structure of this stanza also conforms to the dualistic setup of minemes, with the bird (the exhibit of this part of the text) appearing on both
sides, on one as a metal, on the other as the common, bird.
The exhibit of the image in stanza four is the ame. Its description, spanning
all eight lines, mirrors that of the mouth in lines 1314. On the textual level, the
expression in line 26: Flames that no faggot feeds, nor steel has lit echoes a
mouth that has no moisture and no breath, while thematically, the concept of the
mouth devoid of mouth-ness surfaces in the concept of an un-ame-like ame described in lines 27 and 32: a ame no storm disturbs; a ame that cannot singe a
sleeve. Similarly to the mouth, to the ame is also attributed an autogenetic nature.
The usual sources of re have already been denied in line 26, while line 27 expresses
the notion of autogenesis directly: ames begotten of ame.
The question naturally arises that if all minemes are arranged in the dual structure of the material and eternal worlds, to which side do mouth and re belong?
Since both are deprived of their everyday qualities, and as they are autogenetic, they
have no heterogeneous antecedents, that is to say, they are supposed to have been in
existence since the beginning of time, they appear to belong to the eternal realm.
Furthermore, this placement of re is consistent with its usage in Sailing to Byzantium, where it was aligned with holy-ness, gold and the sages.
Complexities and fury reappear in line 29: complexities of fury leave, presumably eeing from the spirits, which are blood-begotten (28): that is, they
originate from the realm to which man belongs (c.f. veins and blood in lines
68 and 24). It is the re that burns away the material side of man, as complexities of fury are Dying into . . . / An agony of ame (2932). As spirits are contrasted with both complexities and fury, and they originate from the world of
man, they are also presented as elements that bridge the ephemeral and the eternal realms.
Stanza ve furnishes us with one further element bridging the two worlds; its
exhibit, the dolphin. The eternal realm is represented by the now cleansed spirits,
which, however, are astraddle on the dolphins mire and blood (33), that is, on the
material, earthly side of the creatures. The spirits are to be carried across the dolphin-torn, that gong-tormented sea (40). In this stanza, one can also witness the
reappearance of Emperor and gong from stanza one, creating a frame around the
104

SAILING TO BYZANTIUM

poem. While it is as early as the fourth stanza that the element Emperor reappears
along with an important time adverbial it is the fth stanza, where it is paired
with gong, that serves as a counterpart, or a continuation, a closure of the scene
starting in the rst one. This technique may enhance a feeling of formal cohesion,
thus creating a sense of unity in the reader.
Like Sailing to Byzantium, this poem also has a temporal dimension. The rst
line suggests that the time is around dusk, while by line 25, it is already at midnight. According to these elements, the narrative layer of the poem spans a few
hours only - up to the moment usually associated with supernatural events. This
temporal span is more concrete and more closely connected to the representation,
to the images, than the time-span of Sailing to Byzantium, which covers a whole
course of life and is more related to the interpretation.
Since the structure of Byzantium is similar to that of Sailing to Byzantium
in that it also contains a series of almost self-contained images, I will follow the
same strategy of establishing the global exhibit of this poem: namely, by searching
for minemes which are present in most of the images. Similarly to the case of the
previous poem, and somewhat stretching the denition of the exhibit, I suggest that
it is not a single mineme that is the global exhibit of the poem, but a complex of
several ones. Since the elements complexity, fury, mire, vein / blood appear in all
the images (stanzas) except the second, let me assume that in layer II of Byzantium, they occupy the position of the global exhibit.
The properties of the minemes in the two poems show interesting tendencies if
compared to each other. Generally, these elements in Byzantium appear to be
more concrete, that is, more readily able to be visualized than the ones in Sailing to
Byzantium. To illustrate this proposition, let me compare the minemes in the two
bird-images in the two poems instead of attempting to catalogue all the elements in
the texts which can be analysed as minemes. Specically, the comparison is between
the fourth stanza of Sailing to Byzantium, consisting of 8 lines and 58 words, and
the third stanza of Byzantium, containing 8 lines and 47 words. The table below
lists words and short phrases that I consider minemes.
In the analysed stanza of Sailing to Byzantium, 13 words have been selected
as belonging to minemes (either stand-alone or integrated into phrases), whereas
in the other poem, I found 19 words related to minemes, of which one is not a
content word. Taking into account the number of total words in these passages,
the higher percentage of mineme-related words in Byzantium is striking. This
phenomenon might be related to the more nominal style of the latter poem.

105

ELD PL CSIRMAZ

total nr. of words

minemes

nr. of words selected

% of words selected

Sailing to Byzantium
lines 2532
58
nature
drowsy Emperor
lords
ladies
Grecian goldsmiths
hammered gold
gold enamelling
golden bough
13
22

Byzantium
lines 1724
47
bird (2)
cocks of Hades
moon
metal
common bird
petal
mire
blood
golden handiwork (2)
star-lit golden bough
19
40

Minemes in one stanza of Sailing to Byzantium and Byzantium

The sheer number of minemes, however, does not, in itself, guarantee a more effective motivation of images. The quality of the minemes to trigger associations, to be
capable of visualization, their concreteness, is a more important factor. Still, the concreteness of these elements shows the same difference between the poems. To take an
example from the stanzas analysed above, the sub-image of the bough in Sailing to
Byzantium is motivated by the sole adjective golden, whereas in Byzantium, star-lit
can also be found adjoined to it. As suggested above, any further specication makes a
mineme more concrete and more easily able to be visualized; therefore, star-lit golden
bough is the less abstract element. Also, while the scrutinized stanza of Byzantium
does contain a number of abstract minemes (e.g. bird, metal), the mire, petal, cock
elements found among its one-word minemes are more concrete than lords and ladies
or nature found in the stanza in the other poem. Nature is on the borderline of being a
concept, and, as such, conveys only general visual information. It also includes mire,
petal and cock, making it more abstract than those three elements. The phrase lords
and ladies, in turn, refers to a whole social class, and, I think, is closer to its respective
basic object (person) than, for example, cock is to bird, the basic object that it is a part
of. Similar comparisons may show that in the whole text of Byzantium minemes are
more abundant and usually more concrete than in Sailing to Byzantium.
This effect is enhanced further if, when reading Byzantium, one takes the
mythological interpretation of the dolphin as a vehicle or avatar of the soul in transit
into account. The connotations raised by the bobbin bound in mummy-cloth (11) can
106

SAILING TO BYZANTIUM

also enrich a mythologized interpretation. As concreteness is primarily reliant on the

ability of a mineme to trigger associations which enables the reader to construct a
more detailed visual model these mythological associations also suggest the more
elaborated concreteness of the latter poem. Sailing to Byzantium, according to my
reading, contains fewer elements with similar connotations, which is not to say that
it is devoid of them. Like Byzantium, it also refers to the Fates, through perne
aligned with gyre (19); however, this reference might not be immediately interpretable for some readers. Conversely, golden bough (30) does open up a wide range of
possible connections, but this element also appears in the later text, and, therefore,
cannot contribute to the differences between the two poems.
This difference in concreteness also characterizes the exhibits of the stanzas of
the two poems. Setting bird aside, which appears as a local exhibit in both texts,
young, aged man and sages the exhibits in Sailing to Byzantium, which are all
subcategories of the rather general person, specied by mostly one property only
appear to be less concrete than the exhibits of Byzantium: dome (a quite specic
subordinate category of the basic object house), mummy (a subcategory of corpse,
or even the general person), and dolphin (following a nave taxonomy, a subcategory of sh easily capable of visualization). Flame, the exhibit in Byzantium not
mentioned so far, may be at the same level of abstraction as aged man. Moreover,
young, aged man, sages and bird are more connected to the represented interpretation (layer III) than to the representing images (layer II), echoing the difference
between the texts already established, based on their different time-spans. Comparing the global exhibits yields similar results. The global exhibit of Sailing to Byzantium, bird and song, is a complex of two rather abstract minemes if compared to
the minemes occupying this position in Byzantium. Song may be a basic-level
object itself, but as it refers to an auditory phenomenon, it conveys little visual information. Bird is a good example of a basic-level object, and, perhaps more importantly, it contains a large number of subordinate objects (from albatross to
woodpecker) designated by a single word. Turning to Byzantium, I will set two of
the minemes (complexity and fury) serving as its global exhibit aside, as they are
quite abstract, concept-like elements. The remaining ones (mire, vein and blood)
still show a higher level of concreteness than either song or bird. While it is doubtful
whether mire, vein and blood are basic objects or parts of one (perhaps of the objects land, vessel and uid), in neither case does either of them contain as large a
number of subordinate objects as bird does. In other words, while bird only evokes
an averaged example of many, more concrete objects, mire, vein and blood refer to
less varied and therefore more concrete categories.
At the same time, however, the shift of the global exhibit from bird/song to the
fury/mire/vein/blood complex also makes it part of the dual arrangement of min107

ELD PL CSIRMAZ

emes in the later poem. Bird and song in Sailing to Byzantium were elements
which bridged the realms of ephemeral nature and eternal art. The fury/mire/vein/
blood complex, in contrast, belongs undoubtedly to the natural side of this dichotomy. This is not to say that Byzantium lacks bridging elements; in fact, it contains
more than its companion piece. Bird appears on both sides in the third stanza, and
the bridging nature of spirit and dolphin has already been suggested. Also, while in
Sailing to Byzantium the bridging elements bird and song are closely related, in
Byzantium bird stands apart from dolphin and spirit, even if the latter two are
connected via the mythic interpretation. The larger number and separated nature of
bridging elements, in turn, may slightly lessen the effect of the latter poem by obscuring the world model it depicts: transition from ephemeral to eternal is now
possible not only via (Byzantine) art represented by the bird and singing (possibly
signifying the very act of writing), but also via a more mythological route across the
sea. Incidentally, while both the dome of Hagia Sophia and, more importantly, the
dolphin are present in the early drafts of Sailing to Byzantium, Yeats abandoned
them in the writing process.47
The extent of alienation in the two poems can be determined based on the
presence and the positioning of the lyrical I. I have suggested that in Sailing to
Byzantium references to the lyrical I in layer I are more frequent, and their
placement also allowed situating the lyrical I as an aged man, which suggestion
connects the lyrical I with a reconstruction of the biographical author at the time
of the act of writing. Compared to this, the presence of the lyrical I in Byzantium
is extremely limited, and allows no further inference about its position than that of
the addresser of the text. These observations show that alienation is more complete
in Byzantium than in Sailing to Byzantium.
The extent of metathesis in the two poems shows, expectedly, approximately
the same distribution. Representing an artist by a singing bird, as happens in Sailing to Byzantium, does not appear to be a metathesis wide enough to open up an
unprecedentedly wide range of possible interpretations. The feeling that its central
theme, the contrast between art and nature, is rendered into layers II and I in an
unpretentious manner is enhanced by the direct, almost conceptual phrases in the
poem itself. In Byzantium the set-up is not inherently different, but more complicated in the respect that the ephemeral side of the I in layer III is represented by
the more concrete image of the complexities of fury, mire, veins and blood. The
(re)introduction of spirit and dolphin to complement the bridging elements song
and bird, while it can obscure the world-model of the poem, may well serve to pro47. Bradford, p. 115.
108

SAILING TO BYZANTIUM

vide an alternative to the representation of the transition of the artist to that of a

singing bird, a somewhat conventional metaphor.
However, I do not see the main difference between the two Byzantium poems
as one on the level of meaning, interpretation or referents. A basic dissimilarity that
determines many of the textual structures reveals itself if one considers the distribution of minemes in layers A and B. The two similes in Sailing to Byzantium put all
elements related to the city (the mosaic and the machinery) in layer B. In Byzantium it is almost impossible to determine a straightforward descriptive relationship
between images or minemes. Apart from the simile in line 20, according to which
the bird can like the cocks of Hades crow, the minemes are more coordinated than
subordinated to each other as describers and describeds. This arrangement has the
consequence that all images starting from the dome, including the mummy, the
bird, the ames and the dolphin are placed in layer A, in the layer of the view.
That is to say, while Sailing to Byzantium depicted the desired, eternal state in
layer B, Byzantium does this in layer A.
If one interprets the latter poem against the previous one, the fact that this series of images has originally been a vision becomes apparent. Sailing to Byzantium, in other words, allows one to suppose that all elements in Byzantium
originally resided in layer B, and have moved to layer A by the next version of the
text. This is the movement characteristic of symbolism and of myths. Turning to the
biographical author once more, one may note that a piece of writing whose textual
organization is reminiscent of that of myths ts well in Yeatss oeuvre. Moreover,
this afnity of Byzantium to myths may also put in a new light the fact that it is in
this text that more and more direct mythological allusions have been found.
The suggestion that Byzantium, in itself, contains only a covert indication
that what it presents as view has been, in fact, vision, may explain why Byzantium
appears to be a harder, more complex text to many readers. Read without the other
Byzantium poem, it can fail to guide the reader to a reassuring interpretation, placing the images it conjures in relation to one another and in relation to the reader
and the lyrical I. Presenting a vision that is already there, the text offers few
entry points for the reader into the depicted realm.
The titles of the poems may also be related to this observation: in Sailing to
Byzantium, Byzantium, the eternal state, is a mere vision towards which the lyrical
I strives.48 The movement itself guarantees that the reader, identied with the
lyrical I, is guided into the symbolic city. In Byzantium, the lyrical I is already
in the city, as the vision is already there in the view.
48. The rst complete version of Sailing to Byzantium was, signicantly, titled Towards
Byzantium (Bradford, p. 113). Byzantium does not appear to have had an alternative title.
109

ELD PL CSIRMAZ

Conclusion
The analyses of the two Byzantium poems of W. B. Yeats have shown how a common theme may be treated in essentially different ways. Sailing to Byzantium
contrasts the view of the ephemeral, worthless world with the vision of eternal art
and Byzantium; its structure is more linear, with the temporal dimension rooted in
the interpretation, not in the representation, as happens in Byzantium. In the
latter poem, vision takes over, expelling view, and becomes the view itself; the lyrical I is more suppressed; linearity, apart from some vague hints at the passing of
time, almost disappears.
At the same time Byzantium, especially because of the disappearance of the
view, is harder to interpret in itself. It is as if the poem relied on Sailing to Byzantium for a context against which it becomes intelligible.
All of the differences between the poems listed above and taken from different
layers and levels point to the suggestion that Byzantium is an inherently symbolic
text whereas Sailing to Byzantium approximates that mode of writing without
being entirely controlled by it. This conclusion links and puts in a new light the
various observations (regarding the presence of the lyrical I, the use of rhetorical
gures, etc.) made about the two Yeats texts, and it also connects the arrival at symbolism with the arrival in the city of Byzantium, a place constructed entirely out of a
vision that appeared in front of the traveller.

110