The Book of Memory
Page 28
especially since all the numbers of the book- or chapter-divisions are given
in order (as ‘‘Augustinus: ep. 9. 11. 15. 20. 38. 50. 57; De civ. dei li. 10, 11, 12,
13,’’ etc.).147 Why carefully order the one, but not the other, if one were
organizing separate slips of parchment? That the design extends as far as
authors, and then skips, as it were, to chapter numbers (the order in which
a particular work is recalled), suggests that the citations are given in the
order in which Grosseteste had stored them in memory, and not only from
a physically manipulated source. 148
So why was the subject concordance written down (though incom-
pletely)? Presumably to be of use to other scholars, as indeed it proved
for a long time to be. Grosseteste’s tabula amounts to a florilegium without
the texts themselves being written out. A note in the manuscript tells us
that Adam Marsh extended his master’s signs and classifications; indeed,
space is left under some entries in the Lyons manuscript, presumably for
additional entries. This concern for expansion suggests that Grosseteste’s
system became public, presumably when his library came to the Oxford
Franciscans. The index guides a student with access to this particular
library to making his own index by memorizing those passages to which
the notae lead him.
150
The Book of Memory
Another interesting use of alphabetical organization is the second part of
Book II of Albertus Magnus’s treatise, Mineralia. 149 Here the chapter
headings state the nature of the organizing principle: ‘‘De lapidis pretiosis
incipientibus ab A,’’ ‘‘Concerning precious stones beginning with A,’’
and so on through Z. This would seem to reflect the way Albertus had
organized his material mentally. Especially interesting is the fact that
the names are not in absolute alphabetical order under each letter; so
under B there are Balagius, Borax, Beryllus; under C, Carbunculus,
Chalcedonis, Calcaphanos, Ceraurum, Celidonius, Celontes, Cegolites.
Peter of Ravenna’s lists also are not absolutely alphabetical; this suggests
someone arranging material mentally rather than alphabetizing written-
out words with the aid of slips, simply because it is easier mentally to
organize material using a few letters at a time than to alphabetize fully a
number of long words. The latter is better done when one can manipulate
written words physically, as anyone who has had to prepare an index will
recognize. His method suggests strongly that Albertus is using a mental
alphabetical heuristic, which he has simply transposed to a written docu-
ment, writing on the page as he had written in memory before.
One alphabetizing principle that seems very odd and useless to us is
employed in the index to a late thirteenth-century manuscript of
Sentences commentary by the Dominican friar Richard Fishacre, who
died in 1248 (Oxford, New College MS. 112).150 The indexing words are
arranged not according to initial letter but by their vowel combinations;
in other words, by syllables rather than by letters. An explanation of the
system is given (fo. 322):
this preceding index is arranged according to the order of the vowels in the
alphabet, and according to the manner of their combinations [with consonants];
if you wish therefore promptly to find those things which are contained in the
preceding book, take the subject concerning which the passage is made (princi-
pally) or should be made, and regard the vowel of its syllable, or of each of its
syllables if it should be a two-syllable word; and having recourse to the index you
should find the vowel or vowels written in the margin and ordered as previously
said. And opposite the word you will find written the folio number, the page
number, and even the line number in which you will be able to find what you seek;
knowing this also, that ‘‘a’’ designates the first column [of the book opening], ‘‘b’’
the second, ‘‘c’’ the third, and ‘‘d’’ the fourth.
Accordingly, the index begins with monosyllables in ‘‘a’’ (‘‘pax,’’ ‘‘laus,’’
‘‘pars’’); then bisyllables having ‘‘a’’ in each syllable (‘‘adam’’); then ‘‘a’’ plus
‘‘e’’; ‘‘a’’ plus ‘‘i’’; ‘‘a’’ plus ‘‘o’’; ‘‘a’’ plus ‘‘u’’; and so on through all five
vowels singly and in combination. Only the first two syllables of
Elementary memory design
151
multisyllabic words are considered for indexing purposes. This produces a
list in which locutio and pondus both precede locus, pax precedes amor, and
terra precedes spera, for the initial consonants are not considered in the
scheme.
Fishacre’s indexer belonged to a culture that still learned to read Latin by
recognizing syllables. They were the basic units from which words were
thought to be built. Word-recognition and retention was basically a matter
of syllable-recognition. Thomas Bradwardine’s advice for word-memory is
to make an image for each of the five vowels, and then for each possible
combination of one consonant plus vowel. Remembering a whole word
consists of stringing together the various images one has pre-formed for its
syllables. The Fishacre index is based upon a similar understanding of how
words are recalled, simplified to a string of no more than two. Why the
compiler virtually disregarded the order of the consonants accompanying
the vowels is mysterious. Evidently, in his mental arrangements, all the
possible ‘‘a’’ syllables were grouped together, as Albertus Magnus grouped
things by their initial letter. A scheme that groups by syllables is not
inherently less useful as a mental heuristic than one which groups by initial
letter. There were nineteen letters in the Roman alphabet, twenty-one if
one includes ‘‘x’’ and ‘‘z,’’ but these were rarely used as initial consonants.
But the Fishacre index allowed forty-five groups altogether, so that there
are actually many fewer items within each cell of this grid than there are in
initial-letter schemes. It should be noted, however, that no other surviving
index uses this heuristic.
A mental library of dictiones and a library of physical books used the
same heuristics. Having already discussed at length the enduring image of
the trained memory as a library, I will conclude this chapter with a brief
reflection on their parallel cataloguing systems. The cupboard-like press in
which medieval books were kept could be called an armarium, ‘‘cupboard,’’
or a columna, ‘‘column,’’ the word used in a library catalogue from 1400. 151
The books in these medieval arcae or armaria were labeled according to
schemes of letters and numbers, sometimes used separately, sometimes in
tandem. The sources we have for this information are library catalogues
and sometimes the books themselves, mostly from the twelfth century and
later. Typically, each book-press was assigned a letter, and each shelf
(gradus) in it a number, starting from the bottom shelf to allow for
expansion. 152 Sometimes another, subsidiary, nu
mber was assigned to
each volume to indicate its place on its gradus. Often also the lettered
bookcase carried a legend indicating the subject of the contents therein.
Alphabetizing as a heuristic scheme for libraries dates at least to the
152
The Book of Memory
Alexandrine Library. 153 The lay-out of the library and the design of the
scholars’ memories imitate one another. Some of the best evidence for the
similarity medieval scholars perceived between what is read and written in
memory and in books is the way in which heuristic schemes taught for
organizing one’s memorial arcae were also used, when collections became
large enough to need them, to organize the manuscript codices in their
wooden arcae.
C H A P T E R 4
The arts of memory
The mnemonic techniques that I have described so far are basic, their
elementary nature attested by their ubiquity. Divisio and compositio are
processes required by the physiology of memory itself, at least as the pre-
modern world understood it. But it is important to keep in mind that in
considering medieval memoria we are dealing with more than just a set of
techniques or a descriptive psychology, yet also a more specifically realized
value than ghostly ideas (like ‘‘the Gothic’’) which modern students have
attached to some philosophy or other, or to a metahistorical mentalite´.
From antiquity, memoria was fully institutionalized in education, and like
all vital practices it was adapted continuously to circumstances of history.
Memoria unites written with oral transmission, eye with ear, and helps to
account for the highly mixed oral–literate nature of medieval cultures,
which many historians of the subject have remarked. Yet it is also clear that
the later Middle Ages, from the eleventh century onward, was a far more
bookish culture than the earlier medieval centuries had been. Memoria was
adapted to that change, without – as a set of practices – losing its central
place in medieval ethical life. In this chapter I will focus on this change by
considering the revival in the thirteenth century of the Herennian architec-
tural mnemonic which I described in Chapter 2.
The medieval history of the particular mnemonic practice expounded in
Rhetorica ad Herennium and espoused also by Cicero in De oratore is briefly
summarized. Even by the time of Quintilian (first century AD), the method
of projecting images into architectural places and making theatrical scenes
of them, though still known, was considered cumbersome and gimmicky.
Mnemonic practices continued to be cultivated – Quintilian enumerates a
number of them in Book XI of Institutio oratoria – but the elaborate image-
games and exercises to strengthen memoria verborum are not repeated
there. Instead, Quintilian stresses mnemonics for the disposition of one’s
subject matters when speaking ex tempore, and using rebuses and puns to
recall unfamiliar words or especially important items. This practical
153
154
The Book of Memory
attitude persists in the rhetorical teaching on memory of the fourth and
fifth centuries. Indeed there is little evidence of anyone, ancient or medi-
eval, systematically teaching the Rhetorica ad Herennium before the end of
the eleventh century. However, in considering the medieval understanding
of rhetorical craft, including memoria, a historian must clearly distinguish
between those methods described and used in actual practice, and what is
described in and disseminated solely through the two ancient rhetoric
textbooks known in the Middle Ages as the ‘‘vetus rhetorica’’ (Cicero’s
early work De inventione) and the ‘‘rhetorica nova’’ (the Rhetorica ad
Herennium). The medieval titles refer not to the actual order of composi-
tion of these works (for in fact their compositional order is the reverse,
though they were made close in time to one another) but to the order in
which they were incorporated into the medieval curriculum of rhetoric, for
the Rhetorica ad Herennium was not much taught at Bologna, Oxford, or
Paris before the thirteenth century. Its first widely used commentary, and
thus the first evidence of its entering a broadly influential curriculum, is the
Parisian ‘‘Alanus’’ commentary of the late twelfth century. Bolognese
masters of the first part of the thirteenth century, such as Boncompagno
da Signa, report it as little taught, though Boncompagno had clearly read it,
for he parodies its medieval title in his own work of 1235, Rhetorica
novissima. Herennian memoria plays no direct role in the development of
medieval memory craft before the thirteenth century, and when it was
revived in university teaching after 1250, it was within the context of an
extraordinarily rich, diverse understanding of memoria, which had devel-
oped in monastic traditions of ruminative meditation and the composition
of prayer. Monastic memoria is more like what is now called ‘‘mindfulness’’
than what many psychologists deem to be memory, a discipline of attentive
recollection and concentrated reading of texts in the Bible (lectio divina),
which took place during the daily office and in private meditation, and
formed the core of monastic life. But it should not simply be translated as
‘‘mindfulness’’ because the activity of lectio divina assumed a searchable
inventory of remembered materials far richer than any student might be
expected to acquire today. These practices are analyzed historically and
aesthetically in my study of The Craft of Thought, and additionally demon-
strated in the twelfth-century works brought together in The Medieval
Craft of Memory. When the Herennian architectural mnemonic described
by Frances Yates and others was revived in the late medieval universities, it
was within these well-established monastic practices. One cannot separate
it out from them without seriously distorting late medieval and early
modern method and understanding.1 Whether they acknowledged their
The arts of memory
155
debt or not (and, being ‘‘modern,’’ most did not), the Renaissance writers
on the art of memory wrote still from within a profoundly medieval
context. 2
In the late thirteenth century the architectural method enjoyed a revival,
having been commended as the best method by both Albertus Magnus and
Thomas Aquinas. 3 The revival of Tullius’ art of memory occurred in the
context of the classicizing fervor of the early humanists, among whom, for
the purposes of this history, we must include the friars chiefly responsible
for the Aristotelian renaissance that brought back into scholarly circulation
his De anima and its appendices, the Parva naturalia including De memoria
et reminiscentia. Aristotle gave a philosophical explanation for and author-
itative vindication of using arbitrary associations as a basic mnemonic
tool. Both Thomas’s and Albertus’s commentaries on De memoria use
the memory advice of the Ad Herennium as their prime example of the
application of Aristotle’s general precepts concerning the associat
ive nature
of recollective searching. In addition to the Aristotelians, the self-described
Ciceronians of early humanist Italy played an important role in the revival
of the Herennian mnemonic; indeed, some evidence points to the circle
around Brunetto Latini as one agency of this revival. And it is the identi-
fication of the Herennian mnemonic scheme with humanism that led to
its dominance in the memory texts of the Renaissance, many of which
emanate from a milieu that is Italian and also Dominican. To that story,
however, I will return after examining more particularly medieval
schemes. 4
J O H N O F G A R L A N D ( C . 1 2 3 0 ) A N D T H E B E S T I A R Y
The sources of the specific medievalness of the arts of memory I will discuss
now are not attributable to handbooks and manuals, such as some
(unknown) non-Ciceronian body of rhetorical precepts. They are related
instead to manuscript painting conventions, the Bestiary, and various
conventions of pictorial diagrams. In the early Middle Ages, memoria is
discussed not in the context of how best to teach rhetoric, but rather
in writings on meditation and prayer, in which a diagram-like picture
(pictura) is created mentally, which serves as the site for a meditational
collatio, the gathering into one place of the various strands of a meditated
composition. A particularly illuminating and complex example utilizing
the principles of medieval memorial picturing is Hugh of St. Victor’s
treatise ‘‘On constructing Noah’s ark,’’ which I will discuss at length in
Chapter 7. 5
156
The Book of Memory
That text is not, however, an art of memory produced within the
universities, the subject of the present chapter. In particular I will examine
three academic texts which defined ‘‘arts of memory,’’ written by John of
Garland, Albertus Magnus, and Thomas Bradwardine. They all incorpo-
rate a medieval understanding of the precepts of memoria that has wrought
a thorough sea-change upon the Herennian mnemonic. I have chosen to
discuss them out of chronological order because neither Bradwardine nor
John of Garland show any close affinity to the Ad Herennium, but Albertus
does – therefore, John of Garland and Bradwardine are good examples of