Jump to ContentJump to Main Navigation
Contextualizing CassianAristocrats, Asceticism, and Reformation in Fifth-Century Gaul$

Richard J. Goodrich

Print publication date: 2007

Print ISBN-13: 9780199213139

Published to Oxford Scholarship Online: October 2011

DOI: 10.1093/acprof:oso/9780199213139.001.0001

Show Summary Details
Page of

PRINTED FROM OXFORD SCHOLARSHIP ONLINE (www.oxfordscholarship.com). (c) Copyright Oxford University Press, 2016. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a monograph in OSO for personal use (for details see http://www.oxfordscholarship.com/page/privacy-policy). Subscriber: null; date: 05 December 2016

(p.235) Appendix 2 Textual problems in De institutis 3

(p.235) Appendix 2 Textual problems in De institutis 3

Source:
Contextualizing Cassian
Publisher:
Oxford University Press

Cassian's De Institutis 3. 4 poses a problem for liturgical historians that, despite the various explanations tendered over the years, remains, in the words of Robert Taft, ‘the outstanding problem in the history of the formation of the Divine Office’.1 De Institutis 3. 4–6 are, according to another historian, ‘amongst the most problematic texts ever to confront the historian of monasticism’.2 If we were to try to describe the problem in geological terms, these chapters would be an erratic: a large mass of stone transported for miles by a glacier and then dropped in a field where it has no business being after the ice melts.

The erratic in De institutis 3. 4 is a counting error. In De institutis 3. 3, Cassian provides biblical justification for two nocturnal monastic offices and three diurnal offices. The next chapter (3. 4) describes a monastic office that had been added to the cursus in Palestine. According to the writer of this chapter, this new office was not an inappropriate novelty, because it conformed to a line from a psalm of David: ‘seven times in a day have I offered you praise on account of your just judgements’.3 The problem arises from the fact that the five offices described in De institutis 3.3 plus a new Palestinian office only add up to six offices. A variety of ingenious theories have been proposed to reconcile this mistake, but none have won universal acceptance.

One solution that has found little support (and enjoyed even less discussion) is Owen Chadwick's theory that De institutis 3. 4–6 is a later addition to the text.4 Cassian did not write these chapters, but rather they were inserted by a writer who sought to justify liturgical practices that post-dated (p.236) Cassian.5 Cassian actually offered a five-office cursus, rather than the six or seven that most liturgical historians attribute to him. The counting error is simply a clumsy anachronism: six offices were in common use when our later writer added this text, and he simply overlooked the fact that Cassian had elsewhere only prescribed five.

This appendix will demonstrate that Chadwick's solution is both sound and quite likely correct. Moreover, in addition to the three chapters Chadwick singled out as suspect (3. 4–6), there is considerable evidence that Chapter 3. 8 is also a later addition to Cassian's work. A synthesis of traditional textual and new computer-based stylometric methodologies will be employed to probe these questionable chapters. I will begin with a contextual analysis, evaluating these chapters against the larger background of Cassian's work and then probing them for inconsistencies. As we shall discover, the most egregious problem is not arithmetic, but rather that these chapters are devoted to the justification of a period of sleep after Nocturns, a practice that Cassian opposed elsewhere. Moreover, outside of these chapters there is no evidence in Cassian's work for a six- or seven-office cursus, although there is abundant evidence for a five-office cursus.

The contextual argument will then be supported by computer stylometry. Advances in both methodology and reliability have provided the researcher with a powerful new tool for assessing questions of authorship. A stylo-metric assessment of De institutis will provide further evidence that the chapters in question are not consistent with the rest of De institutis 1–4. The cumulative value of the evidence garnered through these two approaches suggests that Chadwick's conclusion about this material is correct. De institutis 3. 4–6 and at least part of 3. 8 are the work of a later hand. Although this might seem a minor point, its significance for the history of the development of liturgy cannot be overstated. Cassian is one of the few reliable witnesses to early liturgical developments. A demonstration that he only knew and recommended a five-office cursus will alter our understanding of the development of monastic liturgy, and will solve ‘the outstanding problem’ in the history of the divine office.

The problem

The stated purpose of De institutis 3. 4 is to add a new office to the five that had been described in the earlier chapters of the work. These five offices (p.237) (Nocturns, Terce, Sext, None, and Vespers) are secure in De institutis. The two nocturnal offices were discussed in De institutis 2. 6; likewise, the three diurnal offices (Terce, Sext, and None) are first specified in De institutis 3. 1. The problem begins in De institutis 3. 4, where a sixth office is proposed. According to the writer, the Gallic monks had adopted a morning office that had been instituted in Palestine. With the addition of this new office (concluded the author of this text), the monks would offer praise to the Lord seven times a day, just like David. Unfortunately, only six offices have been clearly enumerated in the text, despite the writer's claim to the contrary. The problem that has vexed liturgical historians, is trying to decide which of the offices found in later western rules — the Rule of the Master (Regula Magistri) and the Rule of Saint Benedict (Regula Benedicti) — is based on this new office. If the later cursus of offices evolved out of Cassian's recommendations,6 then which of these later offices did Cassian propose (Matins, Prime, or Compline), and where is the missing (seventh) office?

In the second edition of his John Cassian, Owen Chadwick outlined the two positions most scholars have adopted on this question.7 The first option has Cassian proposing Nocturns, Lauds, Prime, Terce, Sext, None, and Vespers (seven offices). Lauds immediately follows the end of Nocturns, a conclusion that might be substantiated by De institutis 3. 4. This proposal has Cassian conflating Lauds and Nocturns in his description of the offices, but counting them separately when tallying them against the verse from the Psalter. Cassian's new office was Prime, the office that signalled the start of the day's work.8 The problem with this proposal, as Chadwick noted, is that one must assume that Cassian indiscriminately used the phrase matutina sollemnitas to refer to both Prime and Lauds.9

A second proposal suggests that Cassian actually wrote about the introduction of Lauds in Bethlehem. The office of Prime was a later innovation, one Cassian knew nothing about. As Chadwick observed, although this solution fits the text better, the explanation still has a problem: the reader is left to find another office. Nocturns, Lauds, Terce, Sext, None, and Vespers only add up to six offices. The advocates for the second proposal have turned to De institutis 4. 19, where Cassian mentions the psalms a monk was supposed to recite before retiring in the evening, and found there a seventh (p.238) office, Compline. Unfortunately this office is not as neatly signposted as its supporters might lead one to believe.10 Moreover, its dislocation from the two books that describe the other six monastic offices is rather curious.

Neither of these explanations is entirely satisfactory. They both require a creative reading of Cassian's text and neither adequately account for the missing office. Chadwick's contribution to the debate was the theory that Chapters 46 were a later addition by an unknown revisionist who was trying to provide a precedent for a liturgical practice that had evolved sometime during the centuries that post-dated Cassian.11 As Chadwick argued, the lack of any manuscripts of De institutis earlier than the ninth century,12 leaves ample time for accretions in the text. If an unknown monk wanted to create justification for a later monastic office, then an insertion into De institutis explaining where the new office originated would lend the imprimatur of antiquity to a new (or local) practice. These chapters would then be used to support the practice of allowing monks to have a short rest after Nocturns.

Chadwick also observed that Book 3 flows more naturally if Chapters 46 are removed from the text. Cassian listed five offices at the end of De institutis 3. 3; in De institutis 3. 7, discussing the differences in penance exacted from those who are late to the diurnal or nocturnal offices, he only mentions three day-offices (Terce, Sext, None) and refers to the nocturnal assemblies (Vespers and Nocturns), again for a total of five offices.

In fact, there are no references to the mysterious offices six and seven outside of these four chapters. Moreover, if we accept this block of text as inauthentic, then the problem of reconciling David's seven prayers with the monastic offices also vanishes. It is a mistake that is easily explained if the author who added these chapters lived at a time when a seven-office cursus was the practice of his monastery. The mistake points to a later period than the one in which Cassian wrote.

Chadwick was not dogmatic about this third proposal, and advanced it with ‘hesitation’.13 He did not view it as ‘the most probable in the prevailing state of the evidence’, but thought it ought to be kept in mind if ‘further evidence of the earlier manuscript tradition should come to light’.14

(p.239) His theory has engendered little debate. The only refutation of it came in the course of a response made by Adalbert de Vogüé to a paper published by Marilyn Dunn, who had argued against de Vogüé's position that the Regula Magistri had been written before the Regula Benedicti.15 Dunn suggested that the case for the precedence of the Master was undermined by the theory that the offices of Prime and Compline were late liturgical developments.16 In order to strengthen her argument, Dunn chose to follow Chadwick's proposal that De institutis 3. 4–6 was a later interpolation and that Cassian only advocated five offices.17

Two years later de Vogüé responded to Dunn's critique of his work. Because his rebuttal was aimed at defending the priority of the Master rather than the question of interpolations in Cassian, his response only addressed the question tangentially. He noted that the idea of a septennium went back to Eusebius who cited it in his exposition of Psalm 118, and consequently provided a precedent for Cassian; moreover, the fact that the passage in Cassian was obscure did not suggest inauthenticity, as obscurity was ‘often the case with Cassian’.18 Cassian's new office is an ‘intentional ambiguity’ which refers to both Prime and Lauds, and a ‘bedtime prayer which is none other than Compline, appears in the following book of Institutes ’.19

This strikes me as very unlikely; Cassian's emphasis on an orderly presentation argues against the idea that De institutis 4. 19 contains the missing (p.240) (seventh) office of Compline. Cassian opened his two-book exposition of the monastic office by stating that he was going to outline the most ancient arrangement (regarding the canonical offices) of the fathers for Castor's new monastery.20 Cassian's goal was standardization. He wrote to replace the variant Gallic practices with one clear cursus of monastic offices. The two proposals which purport to explain Cassian's ‘seven offices’ fly in the face of the goals Cassian had stated, the line to which he closely hewed throughout his exposition of the offices (De institutis 3. 4–6, 8 excepted). If there were two morning offices between Nocturns and Terce, Cassian would have felt bound to separate, explain, and justify them both at length (brevity not being one of Cassian's shortcomings). Similarly, even if Cassian had inadvertently omitted Compline in this book (and bungled his maths) is it likely that he would relegate it to a parenthetical remark in a passage concerned with the time at which the weekly servers hand over their duties to the next group?21 Cassian had devoted the better part of two books to describing the monastic cursus; it is inconceivable that he would have glossed over an office in this manner. This offhand mention of evening psalmody is so obscure that it could just as easily be a reference to the Vespers psalmody, which was described in De institutis 2. In fact de Vogüé offered nothing in his response that had not already been said by Chadwick, aside from accusing Cassian of obscurity and intentional ambiguity. Again, in view of de Vogüé's agenda (defending the priority of the anonymous Master), it is not surprising that Cassian was treated in such a cursory manner, although one might have hoped for a more substantial engagement with Chadwick's proposal.

Contextual issues

One of the main factors underpinning the debate about Cassian's monastic cursus is the fact that both Benedict and the Master advanced an eight-office cursus (Nocturns, Lauds, Prime, Terce, Sext, None, Vespers, and Compline). It is widely assumed that both writers appropriated this structure from Cassian. This is not necessarily the case. To the contrary, when the rules are compared it becomes evident that the structure advocated by the Master and Benedict owe more to the arrangement of The Rule of Saint Basil than to Cassian.22 Basil states that the monk must pray eight times each day: at dawn, the third, sixth, and ninth hours, at the end of the day, the beginning (p.241)

Table 2: Proof texts for the monastic offices

Office

Bas. Reg. fus. 37

Cassian Inst. 3.3

Lauds

Ps. 118: 148

Prime

Ps. 27: 3; 5: 2–3

Terce

Ps. 51: 10–12

Acts 2: 14–18

Sext

Ps. 55: 17; Ps. 91

Acts 10: 13; Col. 2: 15

None

Acts 3: 1 Acts 10: 30;

Acts 3: 1

Vespers

Ps. 4: 4

Ps. 140: 2

Compline

Nocturns/Vigils

Acts 16: 25; Ps. 118: 62

Ps. 118: 147–8

of the night, at midnight, and then again just before dawn. These eight times of prayer correspond to Benedict's eight monastic hours.23 Moreover, both Basil and Benedict designate Vigils as the office that is one beyond the perfect Septies dies, and justify it with an appeal to Psalm 118:62.24 Benedict and Basil also state that Psalm 90 is to be recited at Compline, a practice that Cassian does not mention.25

Cassian's independence from Basil's prescription is demonstrated by the different selection of proof texts used to justify the monastic offices, as indicated in Table 2.

The only clear overlap occurs where both Basil and Cassian use Acts 3:1 as a precedent for None. Psalm 118:148 provides a second point of contact for the two, but they differ in their use of the text; Basil assigns Psalm 118:148 to the office after Nocturns and Cassian employs it as a justification for Noc-turns. Rather than following Basil's structure of proof texts, Cassian grounds his exposition in other patristic writers, appropriating the assignment of prayer times articulated by Clement, Origen, Tertullian, and Cyprian.

Clement of Alexandria was the earliest church writer to provide evidence for set prayers at the third, sixth, and ninth hours of the day.26 As Clement noted, these hours were used by those who would limit their prayers to certain times of day, rather than praying, as the gnostic does, without ceasing.27 While there is no need to see Clement as a direct source for Cassian, it is intriguing to note that Cassian does echo Clement's formulation in his contrast between the less fervent Gallic monks (who need to keep set times of prayer) and the Egyptians, who are able to pray without ceasing.28

(p.242) Origen follows Clement, asserting that while unceasing prayer should be the goal of all, a Christian should at the very least pray three times a day, in accordance with the example of Daniel.29 Here is another point of contact with Cassian, who also uses the example of Daniel as a proof text for this prescription.30 Another similarity can be found in their shared example of Peter praying on the roof at the sixth hour as a precedent for Sext.31

Cassian's justification of the three daily offices exhibits even stronger connections with Tertullian's Concerning Prayer (De oratione). In De oratione 25, Tertullian states that the observance of certain hours of prayer will be profitable for the believer, including those ‘common hours’ which have been deemed more solemn in the Scriptures. These are the third, sixth, and ninth hours. Cassian and Tertullian both use the outpouring of the Holy Spirit to justify the third hour,32 Peter's vision for the sixth,33 and Peter and John praying in the Temple for the ninth hour.34 He also justified the custom of three daily prayers by a reference to Daniel's practice.35

Tertullian noted that the injunction to pray at these times tended to be more of a good idea than a command. It would benefit the believer who followed them as if they were a law (quasi lege).36 Keeping this law would then ensure that the Christian was torn away from the distractions of work or other duties in order to pray at certain times, a sentiment echoed by Cassian who believed that the Gauls needed the structure of daily offices to keep them from drifting away from the duty of prayer.37

Although all of the writers surveyed here deploy similar proof texts to justify the three daily offices, Cyprian's On the Lord's Prayer (De Dominica oratione) contains so many parallels with Cassian's work that it would be remarkable if this work was not the model for Cassian's exposition. De Dominica oratione 32–35, Cyprian's discussion of when a Christian should pray is closer to Cassian than any other text. Both Cyprian and Cassian use Daniel as the justification for three daily offices, the outpouring of the Holy Spirit at the third hour, and Peter's vision at the sixth.38 The prayers of Peter and John at the Temple are omitted by Cyprian.

In addition to the common ground Cassian and Cyprian share with other writers, there are several points where they cite the same texts in isolation from other writers. For instance, Cassian and Cyprian both place Cornelius in prayer at the ninth hour, and both highlight the angelic messenger who (p.243) tells Cornelius that God had accepted the centurion's prayers.39 Another distinctive feature of their works is their development of the Christological justification for prayer. Both writers point to Christ's crucifixion at the sixth hour as justification for prayer at that time,40 and they both develop the redemptive implications of Christ's crucifixion. Cassian states that Christ offered himself as a sacrifice to destroy the sins of the human race, taking on our debt, and thereby achieved victory over the powers and principalities.41 Cyprian noted that Christ washed away our sins with his blood in order to redeem us, and achieved a victory through his passion.42 The two writers ground the significance of the sixth hour in Christ's liberating victory. Cyprian is content to assert that the Lord's Passion stretched from the sixth to the ninth hours, but Cassian presses this point further by detailing how Christ descended into hell in the ninth hour and set the captives free.43

Two conclusions may be drawn from the preceding observations. The first is that despite the fact that Cassian was familiar with Basil's rules,44 he chose to follow the line of argument developed by other ecclesiastical writers when justifying his monastic hours. The second point is that for Cyprian, Tertullian, Clement, and Origen, a Christian prayed three times during the day.

This is an important point; as noted above, the desire to see Cassian as the model for the Regula Benedicti and the Regula magistri, coupled with some muddled information in De institutis 3. 4–8, is all that leads one to look for seven offices in Cassian. In fact, the evidence is much stronger for a five-office cursus.45

This evidence begins with Cassian's first reference to the offices maintained by the Gauls. In De institutis 2. 2, while making an unfavourable (p.244) comment about the great variation to be found in the Gallic observances, Cassian noted that some of them (the Gauls) had thought it good, ‘during the daytime offices of prayers—that is, Terce, Sext, and None—to match the number of psalms to the hour in which the office was rendered’.46 There is no mention in this chapter of a fourth diurnal office. While Cassian's purpose at this point was not to detail these offices, it is intriguing that a morning office is omitted from the list. It would have posed no problem to add the office of Prime if there had been one. This omission is repeated in De institutis 3. 1, where Cassian, having finished his exposition of the nocturnal offices, does take up the task of fleshing out these offices: ‘Now the offices of Terce, Sext, and None, which follow the rule of the monasteries of Palestine and Mesopotamia ought to be discussed by us‘.47 Cassian focuses his attention on the diurnal offices, but again there is no mention of a fourth morning office.

As detailed above, in De institutis 3. 3, Cassian supported Terce, Sext, and None with biblical proof texts. The parallels with Clement, Origen, Tertullian, and Cyprian have already been noted, as has the fact that each of these writers wrote in support of the Christian practice of praying three times each day. Again, if Cassian had a fourth morning office in mind, why did he fail to mention it in any of these places, especially in De institutis 3. 3 where he laid out the biblical support for his cursus?

The other significant point to be drawn from this chapter is the fact that after Cassian justified the three diurnal offices, he retraced his steps and (p.245) offered biblical justification for the two nocturnal offices proposed in De institutis 2. 6. Five offices are substantiated from the Bible at the end of this chapter. Following this marshalling of proof texts, Cassian drew his chapter to a close with a summary justification. The parable of the vineyard owner (Matt. 20: 1–6) was offered to support the five office cursus. According to Cassian, ‘he [the vineyard owner] is described as having assembled them in the first hour of the morning, the time that denotes our morning office, thereafter at the third, the sixth, and after this the ninth, to the latest, the eleventh, by which the hour of lamp-lighting is signified’.48

One problem here might be the fact that Cassian has called the earliest of the offices (Nocturns) a morning office (matutinam nostram sollemnitatem). In describing this office in De institutis 2, he typically employed the adjective night (nocturnus).49 Nevertheless, in this chapter, it is only Cassian's choice of terms that clouds the issue. Although it would have been more convenient if Cassian had labelled the first office a ‘night office’, his language is intended to cement the correspondence with the language found in the parable of the vineyard owner. This flexible word selection is also found in the earlier verses of this chapter where he linked the nocturnal offices to the twice-daily Temple sacrifices.50 Vespers corresponded to the evening sacrifice and was further substantiated by the proposition that Christ instituted the Eucharist in the evening and was himself offered as an evening sacrifice the next day. Note that Cassian used the term evening sacrifice (sacrificium uespertinum) rather loosely here: he had stated (De institutis 3. 3. 3) that Christ had been offered at the sixth hour and had penetrated hell at the ninth hour (3. 3. 6). These precise definitions of time are conflated in 3. 3. 9 under the term evening sacrifice (sacrificium uespertinum). Cassian apparently uses evening as a catch-all term to describe any time after the sixth hour.

The same flexibility of language is found in his justification of Nocturns. This office was subsumed under the category of an evening (uespertinis) office (3. 3. 9), but when he discussed the two offices individually, Nocturns was labelled a morning office: matutina sacrificia (3. 3. 9); matutina sollemnitate (3. 3. 10). The biblical precedent for this office was the fact that the Jews had offered a sacrifice in the morning 3. 3. 9. But just as evening was stretched to include the hours after Sext in the case of Christ's death, so, too, morning was extended to embrace all the hours before the first hour. The text of De institutis 3. 3 makes it clear that this matutina vero sollemnitate (p.246) does not refer to the new morning office established in De institutis 3. 4, but rather is the office of Nocturns that had been presented in De institutis 2. 6.

Having noted that Cassian linked Nocturns to the morning sacrifice in the Jewish Temple, it should come as no surprise that he once again called it a morning office. Surely his point was not to specify a time for the office, but rather to make his analogy work (the vineyard owner went out five times during the course of a day to recruit workers for his harvest). Cassian's five offices (which correspond to the vineyard owner's recruiting trips) are Nocturns, Terce, Sext, None, and Vespers. Sealing this interpretation is the fact that the analogy would be shattered if Cassian had intended the morning office to be Prime, for that would have yielded six offices.

The next three chapters contain the problematic recommendations for a new morning office. Ifthese chapters are skippedinordertopursue the current line of investigation, the next chapter that seems to have come from Cassian's pen is De institutis 3. 7, which contains the penalties meted out to those who come late to the offices. Once again, five offices are listed: Terce, Sext, and None, and the night gatherings (Vespers and Nocturns are implied).51 The monk must arrive before the conclusion of the first psalm during the diurnal offices, or before the conclusion of the second psalm at night if he is to avoid the penalty for tardiness. The interpreter is left with two options at this point: either there are no penalties for late arrival to the new morning office, or, on balance a good deal more likely, Cassian did not prescribe a morning office.

Finally, in De institutis 3. 11, Cassian claims that on Sundays a special concession is granted the monks: the offices of Terce and Sext are conflated and replaced by an Eucharistic Mass that is celebrated before the noon meal. The point of this relaxation is to provide a break from the normal strict observance, so that the monks will look forward to Sundays. Cassian stated that the monks only have one service before lunch, the mass (missa).52 Moreover, he claimed that this single office was the product of merging Terce and Sext. Once again there is no indication of a morning office. If there had been a morning office, the monks would have had two obligations to fulfil before lunch: the morning office and the Eucharistic Mass. Yet Cassian indicated (unam tantummodo missam) that this was not the case.

Chapters 4–6, and 8

As demonstrated in the preceding section, Cassian enumerated a five-office cursus at five different points in De institutis. It has also been noted that (p.247) outside of De institutis 3. 4–6 and 3. 8, there is nothing in Cassian's works that would suggest anything other than a five-office cursus. This section will examine the substance of these dubious chapters. After a brief description of their contents, the chapters will be probed to see if they yield any grounds to suspect their provenance.

As stated above, the addition of a sixth office to the monastic cursus is the subject of De institutis 3. 4. In De institutis 3. 5, the writer draws a contrast between the Bethlehem practice and the current situation in Gaul (where a morning office has also been adopted, following the Bethlehem model). The problem in Gaul, according to the writer, is that after this morning office, the monks were returning to their beds. This showed that the Gauls had failed to understand the raison d'être for this morning office. Prior to its creation, certain monks in Bethlehem had been accustomed to return to bed after the conclusion of Nocturns. Lost in their dreams, they had slumbered until summoned for the next office of prayer (Terce). To counter this tendency to somnolescence, a new morning office had been established. This morning office required the Palestinian monks to arise early and stay out of their beds for the rest of the day. The length of their post-Nocturns rest was sharply circumscribed. This new office had solved the problem in Palestine, but it had lost some of its force when transplanted into Gallic monasteries. Although the Gallic ascetics were rising from bed to celebrate the new morning office, they returned to their rest upon its completion, thereby defeating the office's rationale.53

Moreover, in order to return to bed more quickly, the Gauls were rushing through the office in anticipation of further rest. This was a mistake for reasons that had already been detailed in De institutis.54 The monks who return to sleep either lose the purity they have gained through prayer to the machinations of the Devil, or they will be torpid and sluggish throughout the length of the day. The Egyptians, however, avoided this trap by extending their vigils all the way to dawn, when they began work.

De institutis 3. 6 concludes the case for the morning office. The writer notes that although the eldersinBethlehem had added this office, it was not a novelty because they had not changed the order of psalmody. The hymns used in the morning office were sung by the Egyptians at the end of the Nocturns. These are: Psalms 50, 60, 89, and 148. The writer then makes a comment on secular liturgy, stating that Psalm 50 is also sung in the Italian churches in his day, a practice which he believes was derived from the Bethlehem cursus.

De institutis 3. 7 seems genuine. It simply offers the penances exacted for tardiness at either the diurnalor the nocturnal offices. Oddly, in view of the fact (p.248) that the writer had just written three chapters in support of a new morning office, that office is not listed with Terce, Sext, and None in this chapter.

De institutis 3. 8 expands a reference made in De institutis 3. 4, where the writer had noted the change made ‘especially on those days in which an extremely oppressive weariness was produced in those who celebrated the watches of the evening hours up until the neighbourhood of dawn’.55 In this chapter the writer explains that reference by stipulating the practices that guide the Sabbath vigil. Every Friday evening the monks keep a watch through the night, that ends (in the winter) at the fourth cock crow, so that the monks may return to bed for two hours before rising to celebrate the morning office. This concession is granted so that the monks may take a little sleep and thereby have energy for the work of the following day. The body is otherwise unable to function, and will be overburdened with weariness if it is denied this rest. A period of sleep that is as short as even a single hour will preserve all the good that was won by staying awake through the night. Stretching the vigil all the way to dawn is considered an irrational act. A three-part office is prescribed, and the monks are allowed to sit during the office to alleviate weariness.

The problem of sleep

The provision for a special time of sleep after the office of Nocturns (or an all-night vigil in the case of 3. 8) is the common thread joining De institutis 3. 4–6 and the first part of 3. 8. This provision was contraindicated by the sharp warnings against post-Nocturns sleep found in De institutis 2. 13. These warnings are actually an extension of a theme that closed De institutis 2. 12. Cassian had concluded his discussion of Nocturns by noting that after the prayers were finished, the monks returned to their cells where they did not relax into sleep, but rather remained awake and prayed until dawn. At this time, they began their day's work.56

De institutis 2. 13 opened with a statement of just how serious the issue of sleep was for Cassian: ‘if we desire perfection, then we must agree to diligently observe the same practice’.57 The practice is that of staying awake after Nocturns. Although Cassian had stated in a number of places that he was (p.249) going to water down the stricter Egyptian observance for the weaker Gauls,58 he did not compromise on the provision of a morning rest in De institutis 2.

Two reasons are advanced to support the practice of a post-Nocturns watch. The first is that sleep lowers a monk's defences. The spiritual benefit that a monk gains by rising to celebrate Nocturns may be easily dispersed once he slips back into sleep, the state in which Satan can poison a monk's mind with impure dreams. Consequently, the monk is better off to remain awake after Nocturns, guarding his thoughts against the attacks of a jealous enemy.59

The second reason is that the desire to snatch more sleep is actually a form of spiritual sloth. It makes a monk lazy, and engenders a torpor that will blunt his acuity for the rest of the day. The true monk resists the demands of sleep, just as he fights his other carnal urges. A similar line of reasoning may be found in Basil's directives concerning sleep. The Cappadocian Father saw the desire for excessive sleep as a symptom of spiritual sickness. The monk who was lazy in his devotion to the pursuit of God would inevitably discover sleep stealing up on him.60 A craving for sleep was a sign of spiritual sloth. The soul could make no progress toward God while the body dozed. Sleep was so fundamentally opposed to the monastic vocation that a monk should be grateful when he received the summons from the monk whose duty it was to wake the brothers for prayer.61

Sleep has an interesting (and perhaps under-studied) place in the accounts of Egyptian monasticism. Victory over the need to sleep was as pivotal an ascetic discipline as fasting. The Egyptian sources contain stories about the ascetic battle against sleep, and the greatest of the fathers are portrayed as having limited its claim on their lives. One representative account concerns the famous monk Arsenius. Daniel (his disciple) reported that Arsenius' usual custom was to pass the entire night without sleeping, and when morning came, would say to sleep, ‘Come here wicked servant’ and then sleep for an hour.62

The goal of the monk was an ascent to God, transcending the bodily needs that bound him to the material world. The incredible feats of asceticism described in the sources — the unceasing prayer, prolonged fasting, meagre fare, and sleepless vigils—were designed to demonstrate how completely the masters of asceticism had shifted their lives into the spiritual plane. They were imitators of the angels, those spiritual beings who neither ate nor slept, but rather spent all of their time engaged in the unceasing worship of God.

(p.250) The conflict between sleep and the angelic life was illustrated in Palladius' account of his ascetic instructor, Dorotheus. Palladius claimed that he never saw Dorotheus lay down on a mat to sleep.63 To the contrary, Dorotheus' custom was to stay awake all night, praying and weaving palm ropes. Palladius, wanting to know if this had always been the old man's practice, questioned Dorotheus' other disciples about the master. These men averred that Dorotheus had never voluntarily taken a rest, but slept only when overpowered by drowsiness. Sometimes his treacherous foe would seize him while eating, and food would fall out of his mouth as Dorotheus slipped into an uneasy slumber. On one occasion Palladius tried to convince his master to lay down for a rest, and Dorotheus replied, ‘If you succeed in persuading angels to sleep, then you will also persuade the zealous man.’64

These stories offer a context for the views Cassian expressed about sleep. His recommendations in De institutis 2. 13 were firmly rooted in the Egyptian ethos. Sleep was a barrier to spiritual progress. It was a form of spiritual sloth and represented a dangerous time when the enemy could pollute the unguarded mind. Cassian's identification with this view is also confirmed with a story drawn from De institutis 5. Here he recounts an instance when he was caught sleeping after the evening office (uespertina sollemnitate) by Abba Theodore. ‘Oh John,’ said the old man sadly, ‘How many at this hour are conversing with God and embracing him to themselves and retaining his company? Yet you are cheated out of such great glory, lost in the stupor of sleep.’65 Time lost in sleep was time stolen by Satan, time in which the monk made no spiritual progress. This proposition is also substantiated in Collationes, where Cassian notes that three, or at most, four hours of sleep were all that a monk required.66

The writer of the De institutis 3. 4–6, 8 does not seem to share this view of sleep. To the contrary, sleep was a necessity, something a monk required if he was to function during the day, rather than a seductive pleasure that impeded spiritual growth. Sleep snatched between the end of Nocturns and the beginning of the new morning office was permissible; sleep must (p.251) be taken at the end of the Sabbath vigil if the monk was to avoid torpor and weariness in the next day (advice that is diametrically opposed to what was offered in De institutis 2. 13). Clearly a very different frame of reference undergirds the recommendations of De institutis 3. 4–6, 8.

These disputed chapters contradict Cassian's earlier statements. What is extremely odd, is that when De institutis 3. 5 is examined closely, it appears to be a reworking of De institutis 2. 13, but a reworking that reaches a different conclusion. Certain key phrases in De institutis 3. 5 have simply been copied from De institutis 2, and then redeployed like a cento to support an entirely different view. This becomes evident in the clause that offers the first reason for avoiding morning sleep: ne purificationem nostram confessione supplici et antelucanis orationibus adquisitam.67 This clause marks the start of the dependence on De institutis 2. 13: Prima, ne forte purificationem nostram nocturnis psalmis et orationibus adquisitam.68

The next clause, uel emergens quaedam redundantia umorum naturalium polluat,69 may have been drawn from a later work; Cassian had not written about the problem of the nocturnal emissions to this point, and his fullest discussion of the subject occurs in Collationes 12. 8, where we find the phrase: per soporem caro eius uelut redundantiam superflui umoris expellat, condicionem modumque naturae certissime deprehendet, et ita cum experge-factus inuenerit carnem suam post longa tempora se inscio atque ignorante pollutam.70 It is possible that the writer of this text, familiar with Cassian's later work on nocturnal emissions, thought that this was a danger Cassian would have brought out in his discussion. Nevertheless, Cassian had not mentioned this problem. The inclusion of this statement might actually point to the same sort of anachronistic carelessness that led to a miscounting of offices in the preceding chapter. Of course it is also possible, if these lines are by Cassian, that he is simply anticipating his later discussion.

The next clause (in De institutis 3. 5) lists the illusions stirred up by the Devil as potential polluters of purity during sleep: uel inlusio corrumpat inimici.71 This brings us back to the reasons Cassian had listed for the dangers of sleep in De institutis 2. 13: inuidus inimicusquadam somni inlusione contaminet.72

Another worry, that the restoration of sleep itself can cool spiritual fervour: uel certe intercedens etiam puri ac simplicis somni refectio interrumpat (p.252) spiritus nostri feruorem73 is matched by a phrase from De institutis 2. 13: inter-cedens etiam purus sopor.74 This can lead to a spiritual torpor that will stretch throughout the day: ac tepefactos somni torpore per totum diei spatium inertes deinceps ignauosque traducat,75 another phrase paralleled by segnemque tor-porem inferens menti per totum diei spatium uigorem eius obtundat.76

The next sentence of De institutis 3. 5 follows the thought of 2. 13, although it does not contain the strong verbal parallels found above. Both sentences state that after the completion of Nocturns (missa canonica in this verse, orationum canonicarum in 2. 13), the Egyptians do not return to sleep, but rather prolong their private prayers to daybreak. At this time they begin the day's work. Where Cassian had condemned the practice of returning to bed after Nocturns, the writer of this chapter states that the unlawful practice was to return to bed after the new morning office—the resumption of sleep after Nocturns was fine. No attempt is made to reconcile this new directive with what had been offered in De institutis 2. 13.

Another contradiction of De institutis 2. 13 may be found in De institutis 3. 8. Rather than contributing to torpor, the writer claims that the extra period of sleep is necessary to avoid sluggishness throughout the rest of the day.77 The period of sleep after the long vigil is not a time of spiritual danger, but rather is required if the monk is to function at peak efficiency over the next day. This inconsistency is followed in the next chapter (3. 9), by an argument that would seem to oppose this practice. Here Cassian noted that the Sabbath vigil was observed to commemorate Christ's crucifixion. On the night that Christ was crucified, the distraught disciples watched throughout the entire night, granting no rest of sleep to their eyes.78 Because of this precedent, a special office was appointed to commemorate this night on a weekly basis, and it is kept in the same way, to this day, throughout the east.79 The disciples did not return to their beds after the vigil— they granted no rest of sleep to their eyes. If the monks celebrated the office in the same way, as Cassian asserted that they did, then clearly they would not be allowed to return to their beds at the conclusion of the Sabbath vigil.

Benedict and the Master

(p.253) It was suggested above that modern interpretations of Cassian's cursus are skewed by the fact that both Benedict and the Master recommended an arrangement of eight offices. It was also proposed that the first step in an unbiased examination of Cassian would be to stop viewing Cassian through the window of these later developments and consider the cursus he advanced based on the evidence that may be drawn from his work. This has been the burden of the preceding sections of this appendix.

Nevertheless, there are some interesting observations to be drawn from an examination of the later rules: Benedict, unlike the Master, but following Cassian, does not allow his monks to return to bed after Nocturns. ‘In the time remaining after Vigils, those who need to learn some of the Psalter or readings should study them.’80 Both Benedict and Cassian believed that the period following Vigils was to be employed constructively, not wasted in sleep.

This advice stands in stark contrast to that offered by the anonymous Master, who permitted the monks in his monastery to return to bed after celebrating the office of Matins. After the extra rest offered by this nap, the brothers would be fresh for both work and prayer throughout the remainder of the day. This practice was justified by the example of a certain St Helenus, who was said to have taken a rest after Matins.81

Adalbert de Vogüé's demonstration that Benedict used the Master's rule in formulating his own work has been widely accepted for more than twenty-five years.82 If, as has been suggested, Benedict relied upon the Master, what does his deviation from the Master on the issue of post-Nocturns sleep suggest? One wonders if he might not be looking back to an earlier text, perhaps Cassian's unaltered text, which offered an extensive justification for not returning to bed after Nocturns.

It should also be noted that the disputed chapters in Cassian broadly correspond to the Master's views on sleep (as expressed in the quote above). The brothers should have no qualms about returning to sleep, according to the Master. There is apparently no danger of satanic pollution. Moreover, the extra period of sleep ensures that the monk will be rested for the work of (p.254) the day that follows, a sentiment that corresponds to De institutis 3. 5 and 3. 8, and contradicts De institutis 2. 13.

Another point may be drawn from the Master's legislation: according to the Master, the proper time for the Nocturns office was determined seasonally in relationship to the cockcrow (cantum pullorum).83 Benedict, on the other hand, placed Nocturns at the eighth hour of the night.84 In this, Benedict is much closer to Cassian, who in De institutis 2. 17 stated that the monks are summoned to Nocturns by a monk who remains awake all night, praying and keeping track of the time by the movement of the stars until the appropriate hour arrives. Astronomy, not agriculture, was the basis for starting the office.

A search of all of Cassian's works (De institutis, Collationes, and De incarnatione) for references to roosters reveals that a similar term (gallorum cantum) is used in only four places: three times in De institutis, and once in Collationes. The single use of the term in Collationes comes in a discussion attributed to Abba Theonas, in which the old man ties the beginning of Easter (and the cessation of a fast) to the cock's crow at dawn.85 The three uses of the term in De institutis all occur in the disputed chapters of De institutis 3. In De institutis 3. 5, the writer states that the Egyptians are accustomed to rise, ‘even before the cocks’ crow' (etiam ante gallorum cantum) in order to participate in Nocturns. In De institutis 3. 6, the writer notes that the Nocturns are customarily ended after the cocks' crow (post gallorum cantum). Finally, in De institutis 3. 8, the elders are said to limit the Nocturns to the fourth cocks' crow (quartum gallorum cantum) during the winter months so that the monks can get more sleep.

Cassian's undisputed chapters and Benedict agree in the assignment of the start of Nocturns to an hour of the night. The disputed chapters and the Master both measure time by the cock's crow. Again there is the suggestion of Benedict agreeing with Cassian against the Master. These correspondences between the Regula magistri and Cassian's disputed chapters (against the correspondences between Benedict and the undisputed Cassianic text) provide the basis for an extremely tentative suggestion about the provenance of the changes made to Cassian's work. Perhaps the text was changed in sixth-century Italy, in the same monastic milieu that produced the Master's work.

While speculative, the proposal that De institutis 3. 4–6, 8 were inserted in an Italian monastery also has the merit of making sense of the odd sentence that closes De institutis 3. 6, in which the writer stated that in his day, Psalm 50 was used in all the churches throughout Italy.86 It is not immediately apparent (p.255) why Gallic monks would be persuaded to adopt a practice based on its universal use in Italian churches. On the other hand, this sentence certainly could be read as an unconscious slip by a later writer working in an Italian context, who was trying to provide a precedent for a later Italian practice.87

Suspect words

The presence of a rooster in Cassian's disputed chapters is suspicious. In concluding the contextual analysis of these chapters, it should also be noted that there are two other terms that strike one as anachronistic insertions rather than genuine Cassianic prose. These words are missa and hymnus.

Missa is the most problematic of the pair. In later centuries the church used the word to signify the Eucharistic Mass, but this was a gradual transformation of meaning that had only begun to be adopted in Cassian's time.88 The earliest term the church used for the Eucharist was the ‘Lord’s Supper' (κυρακὸν δϵπὸοὸ).89 In the first century, this ceremony came to be called the Eucharist (ϵὐχαρͺστία).90 In the Latin West, writers used the terms oblatio and sacrificium to describe this event.91 According to Jung-mann, oblatio was the standard name for the mass. This did not begin to change until the sixth century92

Missa from the Latin verb mitto originally meant ‘sending out or sending away’. In late Latin it signified the dismissal from a service.93 This service was not necessarily ecclesiastical as the term was also used to describe people leaving law courts. St Avitus of Vienne (c.500) noted that the missa was pronounced in the churches and in the palaces or praetor's courts when the people were sent away from the event.94

The one time Cassian employs the word in Collationes, it serves as a participle (‘sent’).95 A similar use may be found in the single instance of the word in De incarnatione.96 The limited usage of missa in these two (p.256) lengthy texts would seem to suggest that Cassian did not use the term very often in his writing, but to the contrary, the word occurs ten times in De institutis. In five of these instances, he uses the term in the way just described, to signify dismissal from some event.97

Of the other five occurrences of this word in De institutis, one seems to refer to the Eucharistic Mass, and the other four to monastic offices. In De institutis 3. 11, Cassian states that on Sunday, only one Mass is celebrated before lunch.98 The connection between missa and the Eucharist was not well established by the time Cassian wrote, although Ambrose had used the term in this way.99 Since Cassian was referring to the service where the monks received communion, it may be possible that the term was coming into use at this time.100 Such a usage is rare in late fourth-and early fifth-century writers, but Ambrose does offer a precedent for it.

The other four occurrences of missa in De institutis are very peculiar. It is apparent from the context in which these terms appear that missa is intended to mean office. Significantly, all four of these occurrences are in disputed passages.101 What makes this usage even odder is that sollemnitas is the word Cassian normally uses for office.102 The sudden appearance of the missa in De institutis 3. 4–8 is suspicious. This sense of disquiet is further heightened by the observation that there is another place where missa occurs in De institutis, a place that is almost certainly the work of a later hand: missa appears, meaning office, in some of the chapter headings that have been inserted into the extant manuscripts of De institutis. The chapter title for De institutis 2. 13, for instance, reads Quare post missam nocturnam dormire non oporteat.103 Significantly, the chapter itself does not use the word missa. Nor does De institutis 2. 15, which bears a similar title.104

(p.257) In these titles, which were added by a later hand,105 missa has been inserted as an anachronism. It was not Cassian's term, but one placed in the text by whomever added the headings, writing at a time when the monastic office was called a missa. If this conjecture holds, then the view that De institutis 3. 4–6; 8 are also the work of a later hand is further strengthened by the presence of this word in them.

Another word that may be significant is the word hymnus. Despite the fact that Cassian wrote extensively about the offices and psalmody, the word hymnus only occurs four times in Cassian's works. Two of these occurrences are in the disputed De institutis 3. 6. 2 and 3. 6. 4. Another occurrence is to be found in De institutis 4. 19. The final use of the word is in Collationes 21. 26, where it is used in reference to men (the context suggests that they are not monks) who, upon rising, offer the first fruits of their day to the Lord by singing hymns, praying, or hurrying to church. Although there is an obvious parallel between the worldly men who rise and sing hymns before embarking on the day's business, and monks engaged in the office of Nocturns, Cassian is discussing the monastic offices in this chapter. He uses the word hymnus simply to describe something that Christians sing, rather than as a term that is broadly equivalent to monastic psalmody (as it is used in our disputed chapters). The hymnus plays no part in his other discussions of the monastic offices. Once again, the curious use of a term that would gain currency in later usage is suspicious.

A Statistical investigation of Chadwick's proposal

As noted above, Chadwick's explanation for the problems found in De institutis 3. 4–6 were advanced with some hesitation. He suggested that without a new manuscript find to corroborate his view, the theory would have to remain tenuous. The preceding sections of this appendix have considered the place of these chapters within the overarching context of (p.258) Cassian's work. As has been demonstrated, there are good contextual reasons to doubt the Cassianic authorship of these chapters.

Although there have been no new manuscript finds to cast fresh light on this question, the past thirty-three years have witnessed the development of computer-based, statistical methodologies for the determination of authorship. One of the best of these methods, an application of multivariate statistics, was developed by John Burrows and described in a series of papers in the early 1990s. This method allows a researcher to distinguish between the works of different authors with exquisite accuracy; in the absence of a new manuscript find, this methodology may be employed to shed new light on the question of the disputed chapters in De institutis.

Background

The quest for a statistical methodology to allow researchers to solve questions of authorship dates back to the middle of the nineteenth century. Modern stylometry began with the suggestion of Augustus de Morgan (in 1851) that a measure of the average length of words in a text might be used to discriminate between authors. This average was attained by dividing the total number of characters found in a text by the number of words in the text. While this suggestion seems reasonable (some authors habitually use polysyllabic words while others employ simpler and shorter words), it did not prove to be a credible methodology. Years after making this suggestion, it was discovered that the average word length of texts often varied by genre. Consequently, different works by the same author might demonstrate variations in average word lengths, an observation that invalidated this test for authorship attribution.

The next step forward came in 1938 when statistician G. Udney Yule studied sentence length distributions in the writings of various authors. He discovered that authors tended to employ a consistent mix of sentence lengths, which given a large enough sample, could be quantified. The distribution of sentences provided an authorial fingerprint that could be used to suggest authorship. Yule's methodology was applied to the problem of the anonymous Imitation of Christ. Two authors had been proposed for this work, Thomas à Kempis and Jean Charlier de Gerson. Yule calculated the sentence length distributions for the known works of both authors, and then compared these distributions to the distribution for The Imitation of Christ. Yule concluded that à Kempis was more likely to have been the author of the work than Gerson.106 A further refinement to this (p.259) approach was to be found in W. C. Wake's study of sentence length in Greek authors. Building on the work of Yule, Wake's research revealed that Greek authors tended to write sentences that fell into certain patterns of sentence distributions that could be used to differentiate between authors.107 Unfortunately, this technique was not completely reliable; one of its great problems (especially with unpunctuated classical texts) was the definition of exactly what a sentence was. Moreover, the technique also suffered when making comparisons across different genres.

The next advance in statistical methodology was the landmark study of the Federalist Papers conducted by Moesteller and Wallace. The Federalist Papers are a collection of articles written to develop support for the United States Constitution. These eighty-five essays were published anonymously under the pseudonym ‘Publius’. Two days before his death in a duel with Aaron Burr (1804), Alexander Hamilton left a list that revealed the names of the three authors who had contributed to the collection. The three authors were Alexander Hamilton, John Jay, and James Madison. Several years after Hamilton's death, one of Madison's friends challenged this list, stating that Madison had actually written some of the papers claimed by Hamilton. As a result of this counterclaim, the authorship of twelve of the Federalist Papers was no longer certain, and the question of correct attribution exercised literary critics and historians for the next century.

Moesteller and Wallace decided to use computer stylometry to attack the attribution problem. Their approach was to develop a list of seventy function words that were found in each of the tracts of the Federalist Papers. They defined a function word as one that holds a sentence together: articles, prepositions, pronouns, and other particles. These words are found in all forms of writing; their frequency of use is not likely to vary between genres or works that treat different subjects. From this list of function words, Moesteller and Wallace then identified the words that proved useful in discriminating between Madison and Hamilton's undisputed works. By analysing the relative frequencies of these discriminators in the twelve disputed papers, Moesteller and Wallace demonstrated that Madison had actually written the unattributed papers.108

A revolution in stylometric studies came in 1987, when John Burrows began to publish a series of papers describing a new, multivariate technique (p.260) for authorship analysis.109 Burrows' technique represents the closest that statisticians and textual critics have come to finding the ‘holy grail’ of authorship attribution tools, and is now ‘the standard first port-of-call for attributional problems in stylometry’.110

The fundamental premise underlying the methodologies of Moesteller, Wallace, and Burrows, is the idea that authors tend to use certain words at constant rates. The Burrows Method exploits this tendency in a manner that is more mathematically sophisticated than the method employed by Moes-teller and Wallace. The Burrows Method projects text samples into multidimensional space and groups them by their proximity to one another in this space. A brief illustration of this concept may prove helpful.

For instance, let us suppose that an analyst wished to compare three text samples. Each sample is 1,000 words long. A fairly crude way to judge the similarity between the samples would be to compare the frequency of a single word across the three texts. If the texts were in English, one could count and compare the instances of the word and. Let it be supposed, (somewhat arbitrarily) that and occurs 11 times in the first text sample, 21 times in the second, and 23 times in the third. These three data points can be plotted on a single line as shown in Figure 1. In this univariate representation of the three texts it is evident that text 2 and text 3 are more alike than text 1 and text 2, or text 1 and text 3.

A second variable may be added to the graph. Here the occurrences of the word but will be counted for each text. Again, let us suppose that the word but occurs 15 times in text 1, 30 times in text 2, and 28 times in text 3. This bivariate data can be plotted on two axes of a graph (Figure 2). The frequency of the word and is plotted on the x-axis and the frequency of the word but is placed on the y-axis. Again, by visual inspection, it is clear that text 2 and text 3 are more similar to each other than they are to text 1.

An initial foray into multivariate space is made with the addition of a third word, the. In this case we shall assume that the word the occurs 8, 19, and 27 times, respectively. The frequency of this third word is now plotted on the z-axis of the graph, yielding a representation in three-dimensional space (p.261)

Appendix 2 Textual problems in De institutis 3

Figure 1: Univariate representation of three texts

Appendix 2 Textual problems in De institutis 3

Figure 2: Bivariate representation of three texts

(p.262)
Appendix 2 Textual problems in De institutis 3

Figure 3: Three-dimensional representation of three texts

(Figure 3). Once again, text 2 and text 3 appear more alike in their word usage than either text 1 and 2, or text 1 and 3. Additional words may be added to the list indefinitely; unfortunately, once the number of variables passes beyond three words, the human ability to visualize the additional dimensions fails. Nevertheless, using Euclidean distance formulas, the multivariate distances between points may still be measured mathematically. This is the fundamental principle underlying the Burrows Method: text samples that are similar will be close to one another when projected into multidimensional space.

In order to conduct an analysis of one or more texts using the Burrows Method, the subject texts are divided into blocks of a consistent length.111 In the following examples, each text has been divided into blocks that are 1,000 words long. All of the words in the sample texts are then counted by the computer, and a list of the most common words, sorted in descending order (p.263)

Table 3: The most common words in De institutis,1–4

Word

Number of occurrences

et

287

in

264

non

199

vel

178

ad

160

ut

155

ac

137

quae

115

cum

100

est, sed, quod, qui, per, etiam

99–90

de, si, quam, atque, ab, ne, pro

89–60

a, ita, nec, eius, quoque, ex, hoc, secundum, enim, quidem, haec, se

59–40

his, eum, nos, quibus, quo, velut, usque, esse, id, huius, post, sunt

39–30

scilicet, ea, hac, nisi

29–20

by frequency, is produced. Table 3 shows the fifty most common words in De institutis 1–4.

Assuming that these words are suitable for use in the analysis,112 the computer is then instructed to tally the number of times each of these words occurs in each 1,000-word text block. The resulting data matrix is standardized for each word in the list (converting the frequency scores for each word into standard units with a mean of zero), and a principal components analysis is conducted on the table.113 The two primary principal (p.264)

Appendix 2 Textual problems in De institutis 3

Figure 4: Cass. Inst. 1–4 compared to Hier. Vit. Hil.

components are extracted, and each text block is plotted on a two-dimensional graph.114

The next series of charts demonstrate the results of the Burrows Method. In Figure 4, Cassian's De institutis 1–4 has been plotted with Jerome's Vita Hilarionis.

The points that represent 1,000-word segments of De institutis appear on the left side of the chart, while the points that correspond to segments of (p.265)

Appendix 2 Textual problems in De institutis 3

Figure 5: Cass. Inst. 1–4 compared to Sulp.-Sev. Dial.

Vita Hilarionis are found on the right side of the plot. A similar effect is produced when Sulpicius Severus' Dialogi (Figure 5) is compared to Cassian.

Figure 6 offers an example of all three of these texts (Jerome, Cassian, and Sulpicius Severus) processed together. Once again, the works by these three authors have been separated into discrete regions.

Each of these charts was processed in an identical manner, and each demonstrate that the Burrows Method is able to separate texts by different authors, based on the relative frequencies of the fifty most common words in the texts.115

As we have seen in the preceding consideration of the theoretical basis for the Burrows method, texts by different authors can be separated based on (p.266)

Appendix 2 Textual problems in De institutis 3

Figure 6: Cass. Inst. 1–4 compared to Hier. Vit. Hil. and Sulp.-Sev. Dial.

the relative frequencies of one, two, or three words. The Burrows Method groups texts based on their similarity across a fifty-dimensional spectrum. This space is then reduced through principal components analysis to create the two-dimensional charts seen here. The points plotted on the chart are the product of two equations which consist of fifty variables representing individual words and their weighted coefficients. The coefficients of these equations may also be used to produce a scaled loading chart, a graph that shows which words are significant discriminators between various texts.

Figure 7 is a scaled loading chart for the two texts that were analysed in Figure 4. A comparison of this chart to the one examined earlier (Figure 4) reveals that the points on both charts produce similar contours. The scaled loading chart allows the analyst to identify the words that make the largest contribution to the separation between authors found on the Burrows Chart. The words that fall on the extreme right and left edges of the word clusters are greater contributors to the separation: that is, the principal components analysis has assigned a greater weight to them when producing the two equations that map the fifty-dimensional space onto two dimensions.

The value of the scaled loading chart is that it allows us to make some observations about the way Jerome and Cassian write. For instance, in (p.267)

Appendix 2 Textual problems in De institutis 3

Figure 7: Scaled loading chart for Cass. Inst. 1–4 compared to Hier. Vit. Hil.

Figure 7, the conjunction ac is found on the extreme left edge of the word cluster. On the right edge, we find the word et. Both authors use these conjunctions in their writings, but the scaled loading chart suggests that they use them at different rates. In fact, in the texts chosen for analysis, Cassian seems to prefer ac while Jerome regularly utilizes et. This is an important stylistic difference between the two writers, one that is revealed by the scaled loading chart.

De institutis 3

Having established the usefulness of the Burrows Method in separating texts, attention may now be focused on Cassian's work. Figure 8 shows a plot for Cassian, De institutis 1–4 by itself. Most of the data points are grouped in a cluster that is centred in the middle of the chart. Two points, however, stand as outliers to this main group. One point (labelled ‘a’) is located near the top of the chart; the second outlier (labelled ‘b’) is found near the left edge of the chart.

The first outlier (point a) is the text block that runs from De institutis 4. 34 to the end of Book 4. This section contains a speech that was delivered by Abba Pinufius on the occasion of the reception of a young novice into (p.268)

Appendix 2 Textual problems in De institutis 3

Figure 8: Cass. Inst. 1–4

Pinufius' monastery. Two points need to be made about this text block. The first point is that variation along the x-axis of this chart (the horizontal axis that displays the primary Eigenvector) is more significant than variation along the y-axis (which records the secondary Eigenvector). The primary Eigenvector in a Burrows Chart is always the Eigenvector that produces the largest range of data variation. Consequently, point a is not as significant a variation as point b. Nevertheless, there is still some variation there, and one wonders if this was not related to a change in an author's style when trying to reproduce a speech.116

The variation displayed in the case of point b is not explained as easily. This data segment consists of 1,000 words that begin near the end of De institutis 3. 3 and run to the beginning of 3. 8. These are the chapters that this appendix has argued are later interpolations. What the Burrows Method suggests is that there is something quite different about these suspect chapters. They are isolated from the main cluster, a trait that has been shown (above) to signal authorship differences.

(p.269)

Appendix 2 Textual problems in De institutis 3

Figure 9: Scaled loading chart for Cass. Inst. 1–4

Of course this is not in any sense proof that these chapters are by a different author. The Burrows Method is nothing more than an indicator of variations in word usage over a large subject array. Nevertheless, it is striking that the analysis by the Burrows Method supports the more traditional textual analysis made in the preceding sections. These chapters do display a marked and demonstrable variation from the other chapters in De institutis.

The differences between point b and the rest of De institutis 1–4 can be investigated with the scaled loading plot. Figure 9 shows the contributions each word makes to the separation found in De institutis 1–4.

At the left edge of the plot (corresponding to the displacement noted for Point 6), are five words that are making a large contribution to this separation: the prepositions a, in, and post, and the adverbs quoque and usque. Table 4 compares the frequencies of these words in De institutis 3. 4–8 to the rest of De institutis, as well as to Cassian's other works.

This table highlights Cassian's use of these five words in his works. Column 1 contains the word and column 2 lists the number of times this word occurs in De institutis 3. 4–8.117 Columns 3–5 contain averages for the number of times each word occurs (per 1,000 words) in De institutis, Collationes, and De (p.270)

Table 4: Frequencies of selected words (per 1,000 words)

Word

Inst. 3.4–8

Inst. Ave.

Coll. Ave.

De Inc. Ave.

All works Ave.

Min.

Max.

Std. Dev.

a

6

3.55

4.30

5.41

4.34

1

7

1.77

in

24

17.77

16.89

25.82

18.65

10

31

4.90

quoque

5

3.46

2.82

1.59

2.74

0

7

1.77

post

9

1.45

0.98

1.35

1.14

0

6

1.80

usque

11

1.12

0.46

0.15

0.54

0

3

1.92

incarnatione. Column 6 contains an average for all of Cassian's works. Columns 7 and 8 contain the range for the word in De institutis (the fewest and greatest number of times the word is used in a block of 1,000 words).118 The final column contains the standard deviation for each word in De institutis,119 a measure of the spread of the data observations around the mean.

Although the scaled loading plot has indicated that the words a, in, and quoque are large contributors to the variation shown on the chart, in the case of De institutis 3. 4–8, the values are not that extreme. In a normal distribution, it would be expected that 84 per cent of all the data points would fall within 1 Z score of the mean. The first three words in Table 4 all have Z scores of less than 2.0120 Each of the readings could be attributed to the random variation that occurs in a normal distribution.

The same cannot be said about the preposition post and the adverb usque. The Z score for post is 4.19, and the Z score for usque is 5.15. Both of these values would be termed statistically significant, and it is extraordinarily unlikely that they are the product ofsimple random variation. This observation is strengthened by an examination of the range for these words. Within De institutis, post never occurs more than 6 times in a 1,000-word block. On average, post only occurs 1.45 times every 1,000 words in De institutis, 0.98 times for every 1,000 words in Collationes, and 1.35 times in every 1,000 words in De incarnatione. These statistics suggest that Cassian was not in the habit of using post in his writing. Yet, in the disputed chapters, the word occurs 9 times.

(p.271) The adverb usque demonstrates an even more dramatic quality. Outside of our disputed chapters, this word never occurs more than 3 times in a 1,000-word block in any of Cassian's writings. It occurs, on average, 1.12 times per 1,000-word block in De institutis, 0.46 times for every 1,000 words in Collationes, and 0.15 times in every 1,000 words in De incarnatione. Yet in the 1,017-word block that covers De institutis 3. 4–8, the word occurs 11 times. The Z score of 5.15 suggests that this text sample was not drawn from the same population that produced the other text samples.

If one is going to argue that Cassian wrote De institutis 3. 4–6, 8, then some explanation will have to be offered for the unusual frequency of usque and post in these chapters. In all of Cassian's other works, he never used usque more than 3 times in any given 1,000-word block of text, and on average, he only used the word once in every 2,000 words (or 0.54 times per 1,000 words). Based on his normal usage of the word usque, it is extremely unlikely that Cassian wrote these chapters.

Although the unprecedented density of usque and post tend to heighten the feeling that something is not quite right with these chapters, it must be noted that the Burrows Method does not depend on the frequency of these two words alone. In fact, if usque and post are removed from consideration (ignored as if they were context-sensitive nouns) the suspect chapters still separate from the main cluster, as shown in Figure 10.

Appendix 2 Textual problems in De institutis 3

Figure 10: Cass. Inst. 1–4 omitting usque and post

(p.272)

Appendix 2 Textual problems in De institutis 3

Figure 11: Scaled loading chart for Cass. Inst. 1–4 omitting usque and post

As the scaled loading chart (Figure 11) demonstrates, even when the effects of usque and post are factored out, the variations in usage of other words continue to separate the suspect chapters from the main body of De institutis 1–4. In this case, the influence of a, in, and quoque pull the texts apart, as does the relative absence of words that Cassian normally uses, such as solum and uelut. While the differences in usque and post are the dramatic variations, these chapters appear as an outlier to the main cluster because they vary across a wide spectrum of words. The Burrows Plot has demonstrated that the chapters that make up De institutis 3. 4–8 are statistically different from the rest of De institutis.

The apposite judgement of David Holmes bears repeating at this juncture:

The evidence brought forward here should not be regarded as superseding that of the more traditional kind. In attribution of authorship, stylometric evidence must be weighed in the balance along with that provided by more conventional scholarship. Stylometry does, however, have a role to play despite the suspicions of those who mistrust the application of statistical and computing techniques to literature and the analysis of texts.

The way forward in problems of authorship lies in a combination of statistical techniques with more orthodox methods. If the computer reveals unusual, (p.273) quantifiable properties in a text, it is for the scholar in the field concerned to identify the features which are producing these effects.121

The approach to Cassian's De institutis 3. 4–6, 8 in this appendix has been to combine a stylometric textual analysis with an analysis of the passage using more traditional methods. It has been argued that these chapters do not fit into the contextual background for them supplied by the surrounding text. Cassian had argued emphatically against monks returning to sleep after the end of Nocturns. This view had been explicitly expressed in De institutis 2, and was entirely consistent with the teachings of the Desert Fathers reported in other sources. Nevertheless, the writer of De institutis 3. 4–6, 8 stated that there was nothing wrong with this practice, and indeed the monk must return to bed after the Sabbath Vigil if he was to remain awake the next day.

Cassian was also very consistent in his tallying of the offices that made up the monastic cursus. At five different points (including immediately before and after the disputed chapters) he listed five offices of prayer. Nowhere, outside of De institutis 3. 4–6, 8, can a certain reference to a sixth or seventh office be found.

In this context, the use of the word missa is an anachronism, and the suspicion engendered by this word's appearance in the questionable chapters is heightened when one notes that it also occurs (as a reference to the monastic office) in titles that were inserted at a later date into De institutis. A possible connection of these chapters with the Italian monastic milieu of the sixth-century Master has also been suggested; moreover, it has been observed that at certain points (most notably in the issue of a return to sleep after Nocturns) Benedict and the undisputed chapters of Cassian agree against the Master.

And finally, an analysis using a proven statistical method has disclosed the fact that these chapters are demonstrably different in terms of word usage from the rest of De institutis 1–4. The author of these passages relies heavily on the terms usque and post, words which are rarely found in any of Cassian's other writings. When these words are removed from consideration, these chapters still exhibit variation, which suggests that it is unlikely that they were written by John Cassian.

The collective mass of these observations provide a firm foundation for doubting the Cassianic authorship of these chapters. While this contention may not be proved with complete certainty, enough objections to the text have been proffered to substantiate grounds for doubt. Chadwick's ‘hesitation’ about his theory, while judicious, seems less necessary in the face of this new research.

Notes:

(1) Robert Taft, The Liturgy of the Hours in East and West: The Origins of the Divine Office and Its Meaning for Today (Collegeville, MN: Liturgical Press, 1986), 191. Taft's emphasis.

(2) Marilyn Dunn, ‘Mastering Benedict: Monastic Rules and Their Authors in the Early Medieval West,’ English Historical Review 105 (1990), 577.

(3) Cassian, Inst. 3. 4. 3 [SC 109: 104. 39–40] (citing Ps. 118: 164): Septies in die laudem dixi tibi, super iudicia iustitiae tuae.

(4) See Owen Chadwick, John Cassian (Cambridge: Cambridge University Press, 1968), 76–7, and the discussion that follows.

(5) For a good overview of the practice of medieval forgery, see Elizabeth A. R. Brown, ‘Falsitas pia sive reprehensilis: Medieval Forgers and Their Intentions,’ MGH Schriften 33.1 (1988), 101–19. Nor was this practice strictly limited to the mediaeval period (see Eugene Rice, Saint Jerome in the Renaissance (Baltimore and London: Johns Hopkins University Press, 1985), 45–6 for an account of writers attributing their works to Jerome).

(6) Adalbert de Vogüé and Jean Neufville, La Règle de Saint Benoît, Sources Chré-tiennes 181 (Paris: Cerf, 1971), 101–3.

(7) Chadwick, John Cassian, 2nd ed., 73–6.

(8) Owen Chadwick, ‘The Origins of Prime,’ JTS 49 (1948), 178–82, originally backed this position against the work of Dom Jacques Froger. By 1968 he had significantly revised his view.

(9) Chadwick, John Cassian, 2nd ed., 74.

(10) Contra Adalbert de Vogüé, ’The Master and St Benedict: A Reply to Marilyn Dunn,’ English Historical Review 107 (1992), 101.

(11) Chadwick, John Cassian, 2nd ed., 76–7.

(12) We actually do have a sixth-century palimpsest, (Codex F—IV—1 N.16), but this only has fragments of books 4, 6, 7, and 8.

(13) Chadwick, John Cassian, 2nd ed., 76.

(14) Chadwick, John Cassian, 2nd ed., 76–7.

(15) See Dunn, Mastering Benedict, 567–93, and Vogüé, The Master, 95–103.

(16) Adalbert de Vogüé has argued that the Master composed his rule c.500–25 (Dunn, Mastering Benedict, 579). Benedict then used this rule in writing his own rule (c.530–50). According to Dunn, Prime does not appear until Caesarius, who wrote his rule in 534, and Compline first appeared in Italy ‘in the 540s or 550s’ (Dunn, Mastering Benedict, 578–80). Consequently, the Regula Magistri could not have the early date proposed by de Vogüé.

(17) Dunn, Mastering Benedict, 577–8.

(18) Vogüé, The Master, 100–1.

(19) Vogüé, The Master, 101. The view that Cassian was an unsystematic writer has been advanced by a number of different writers, including Peter Munz, ‘John Cassian,’ JEH 11 (1960), 1, and Philip Rousseau, ‘Cassian, Contemplation and the Coenobitic Life,’ JTS n.s. 26 (1975), 113, (although Rousseau softened this by defining unsystematic as allowing for an evolution of thought). Chadwick noted that he had once thought Cassian was very unsystematic, ‘as unsystematic as is possible for the architect of a system,’ but had revised his opinion to suspect interpolations and rearrangements of Cassian’s works by later copyists (Chadwick, John Cassian, 2nd ed., 43). Somewhat ironically, de Vogüé argued for the systematic quality of Cassian's exposition in his analysis of the structure of Collationes (Adalbert de Vogüé, ‘Pour comprendre Cassien: Un survol des Conférences,’ Collectanea cisterciensia 39 (1979), 250–72), and Lauren Pristas, ‘The Unity of Composition in Book V of Cassian’s De institutis,’ SPAT 25 (1993), 438–43, argued that Chadwick's characterization of De institutis 5 as disordered was not the case but rather reflected a highly structured arrangement.

(20) Cassian, Inst. 2. 2.

(21) Cassian, Inst. 4. 19 [SC 109: 146. 19–21]: conuenientibus in unum fratribus ad concinendos psalmos, quos quieturi ex more decantant.

(22) Bas. Reg. fus. 37.

(23) Ben. Reg. 16.

(24) Bas. Reg. fus. 37; Ben. Reg. 16.

(25) Bas. Reg. fus. 37; Ben. Reg. 18.

(26) Clem. Str. 7. 7. Cf. Taft, Liturgy of the Hours, 14.

(27) Clem. Str. 7. 7.

(28) Cassian, Inst. 3. 2.

(29) Or. Or. 32.

(30) Cassian, Inst. 3. 3. 1.

(31) Cassian, Inst. 3. 3. 4; Or. Or. 32.

(32) Cassian, Inst. 3. 3. 2; Tert. Or. 25.

(33) Cassian, Inst. 3. 3. 4; Tert. Or. 25.

(34) Cassian, Inst. 3. 3. 7; Tert. Or. 25.

(35) Cassian, Inst. 3. 3. 1; Tert. Or. 25.

(36) Tert. Or. 25.

(37) Cassian, Inst. 3. 2.

(38) Cassian, Inst. 3. 3. 1, 2, 4; Cypr. Dom. orat. 34.

(39) Cassian, Inst. 3. 3. 7; Cypr. Dom. orat. 33. It should be noted that Cyprian uses the example of Cornelius in a slightly different manner than Cassian: whereas Cassian stresses the time Cornelius received his vision (during prayer at the ninth hour, thus a justification for this hour of prayer), Cyprian offers Cornelius as an example of someone who offered an effectual prayer (which happened to be at the ninth hour). Cyprian's point is that the believer may not pray in a distracted manner, but being in the presence of God, must focus on his task (a theme later developed by Bas. Reg. br. 201). This discussion of effectual prayer then leads directly into the hours for prayer, so if we are making a case for dependence, we could argue that it is there already in Cyprian, even though Cassian employed the verse in a slightly different manner.

(40) Cassian, Inst. 3. 3. 3; Cypr. Dom. orat. 34.

(41) Cassian, Inst. 3. 3. 3.

(42) Cypr. Dom. orat. 34.

(43) Cassian, Inst. 3. 3. 3.

(44) At least Rufinus' translation/codification of the work. It is intriguing to note that Basil's description of the offices (Bas. Reg. fus. 37) was not translated by Rufinus. One wonders if the only text Cassian knew was the Rufinian translation.

(45) Cf. Hier. Ep. 22. 37 [CSEL 54: 201. 11–14], in which Jerome notes that ‘everyone knows that the set hours for prayer are at the third, sixth, and ninth hours, at dawn and in the evening’ (horam tertiam, sextam, nonam, diluculum quoque et uesperam nemo, qui nesciat). Other than moving Nocturns forward to dawn, Jerome's description overlaps perfectly with Cassian's recommendations. Unfortunately, Jerome is not consistent in his recommendations. A later letter (Hier. Ep. 107. 9 [CSEL 55: 300. 17–20]) to Laeta instructing her on how to raise her daughter as a holy virgin) does seem to have a six-office cursus in mind: ‘She ought to become accustomed to rise in the night for prayer and psalms, to sing hymns in the morning, at Terce, Sext, and None to stand in the battle line as one of Christ’s warriors, and when the time to light the lamp comes, to render the evening sacrifice' (et adsuescat exemplo ad orationem et Psalmos nocte consurgere, mane hymnos canere, tertia, sexta, nona hora quasi bella-tricem Christi stare in acie, accensaque lucernula reddere sacrificium uerspertinum). This recommendation is substantiated in the panegyric Jerome writes about Paula (Hier. Ep. 108. 19 [CSEL 55: 335. 7–9]). Here, describing the offices kept in Paula's monastery he gives the following cursus: ‘In the morning, at the third, the sixth, and the ninth hour, in the evening, and in the middle of the night, they were singing through the Psalter’ (mane, hora tertia, sexta, nona, uespera, noctis medio per ordinem Psalterium canebant). Similar advice is given to Demetrias in Hier. Ep. 130. 15.

(46) Cassian, Inst. 2. 2. 2 [SC 109: 58. 2. 12–14]: Sunt quibus in ipsis quoque diurnis orationum officiis, id est tertia, sext, nonaque id uisum est.

(47) Cassian, Inst. 3. 1 [SC 109: 92. 1. 3–6]: Nunc de sollemnitatibus tertiae, sextae nonaeque secundum regulam monasteriorum Palaestinae uel Mesopotamiae nobis est disserendum.

(48) Cassian, Inst. 3. 3. 11 [SC 109: 102. 110–14]: Ita enim et ille primo mane conduxisse describitur, quod tempus designat matutinam nostram sollemnitatem, dein tertia, inde sexta, post haec nona, ad extremum undecima, in qua lucernaris hora signatur.

(49) See, for instance, Cassian, Inst. 2. 4; Cassian, Inst. 2.6; Cassian, Inst. 2. 13

(50) Cassian, Inst. 3. 3. 9; Num. 28: 4.

(51) Terce, Sext, None: Cassian, Inst. 3. 7. 1 [SC 109: 108. 1]; nocturnal offices: Cassian, Inst. 3. 7. 2 [SC 109: 108. 11–12].

(52) Cassian, Inst. 3. 11 [SC 109: 116. 2]: unam tantummodo missam. On the problems of the missa, see the discussion below.

(53) Cassian, Inst. 3. 5.

(54) A reference to Cassian, Inst. 2. 13.

(55) Cassian, Inst. 3. 4. 2 [SC 109: 104. 20–2]: in his praesertim diebus, quibus a uespertinis horis excubias usque ad aurorae uiciniam celebrantibus nascebatur onerosior lassitudo.

(56) Cassian, Inst. 2. 12. 3.

(57) Cassian, Inst. 2. 13. 1 [SC 109: 82. 4–5]: Quod nos quoque, si perfectioni studemus, eadem diligentia conuenit obseruare.

(58) Indeed, the addition of three diurnal offices represents a modification for Gauls of the more pristine two-office Egyptian system (Cassian, Inst. 3. 2).

(59) Cassian, Inst. 2. 13. 1–2.

(60) Bas. Reg. br. 32.

(61) Bas. Reg. br. 43.

(62) Apophth. Patr. Arsenius 14. Arsenius also asserts that a monk only requires an hour of sleep every night if ‘he is a good fighter’ (Apophth. Patr. Arsenius 15).

(63) See also Apophth. Patr. Bessarion 6, where the old man claims to have slept standing or sitting for fourteen years.

(64) Pall. Hist. Laus. 2. 3 [Butler (1904): 18. 1–2]: 'Eàv πέσης τυς ἀγγέλους κοιμηθῆνι, πϵίσις καἰ τὸν σπουδαῖν. See also Apophth. Patr. Poemon 185, where the old man claims that he cannot do without food, clothing, or sleep, but he can restrict his dependence on these things. A similar emphasis on limiting the bodily need for sleep may be found in Apophth. Patr. Sarmatas 3.

(65) Cassian, Inst. 5. 35 [SC 109: 246. 7–10]: quanti, inquit, o Iohannes, hora hacDeo conloquuntur eumque in semet ipsis amplectuntur ac retinent: et tu fraudaris tanto lumine, inerti sopore resolutus?

(66) Cassian, Coll. 12. 15; Cassian, Coll. 13. 6.

(67) Cassian, Inst. 3. 5. 1 [SC 109: 106. 10–11].

(68) Cassian, Inst. 2. 13 [SC 109: 82. 6–7].

(69) Cassian, Inst. 3. 5. 1 [SC 109: 106. 11–12].

(70) Cassian, Coll. 12. 8 [SC 54: 135. 9–13].

(71) Cassian, Inst. 3. 5. 1 [SC 109: 106. 12–3].

(72) Cassian, Inst. 2. 13. 1 [SC 109: 82. 7–9].

(73) Cassian, Inst. 3. 5. 1 [SC 109: 106. 13–15]

(74) Cassian, Inst. 2. 13. 3 [SC 109: 82. 20].

(75) Cassian, Inst. 3. 5. 1 [SC 109: 106. 15–16].

(76) Cassian, Inst. 2. 13. 3 [SC 109: 82.21–3].

(77) Cassian, Inst. 3. 8. 1.

(78) Cassian, Inst. 3. 9. 1 [SC 109: 112. 6–7]: nullatenus quietis somnum suis oculis indulgentes.

(79) Cassian, Inst. 3. 9. 1 [SC 109: 112. 9–10]: in hodiernum diem per uniuersum Orientem similiter obseruatur.

(80) Ben. Reg. 8. 3 [SC 182: 508. 4–6]: Quod uero restat post uigilias a fratribus qui psalterii uel lectionum aliquid indigent meditationi inseruiatur.

(81) Reg. Mag. 33.

(82) Although not universally accepted. See the objections raised in: Dunn, Mastering Benedict, 567–93, and Marilyn Dunn, ‘The Master and Saint Benedict: A Rejoinder,’ English Historical Review 107 (1992), 104–11, as well as the response to these arguments in Vogüé, The Master, 95–103.

(83) The cock's crow is mentioned several times in Reg. Mag. 33.

(84) Ben. Reg. 8.

(85) Cassian, Coll. 21. 25. 3.

(86) Cassian, Inst. 3. 6 [SC 109: 108. 13–15]: Denique per Italiam hodieque consummatis matutinis hymnis quinquagensimus psalmus in uniuersis ecclesiis canitur.

(87) The same sort of slip has already been noted in the numbering of the monastic offices.

(88) For the connection between missa and the Eucharistic Mass, see, Joseph A. Jungmann, The Mass of the Roman Rite (London, 1959), 129–33; F. Brunner, ‘Roman Mass,’ The New Catholic Encyclopedia 9 (1967), 414.

(89) 1 Ep.Cor. 11: 20 [UBS 4: 592].

(90) See, for instance, Did. 9. 1.

(91) Jungmann, Mass, 130.

(92) Jungmann, Mass, 130.

(93) Isid. Orig. 6. 19. 4. Egeria (Pereg. 24. 3) describes the end of the mixed morning service in Constantinople (monks, virgins, and laity) as a missa.

(94) Alcim. Ep. 1.

(95) Cassian, Coll. 18. 16 [SC 64: 36. 19].

(96) Cassian, Incarn. 7. 11 [CSEL 17: 367. 10]: ‘Having been sent by God’ (a deo missa).

(97) In Cassian's classic story on vainglory, the elder claims that he had arrived just as the younger monk was dismissing his imaginary catechumens modo, inquiens, ueni, quando tu missam catechumenis celebrabas (Cassian, Inst.11. 16 [SC 109: 442. 17–18] [two instances in this chapter]). Also in Cassian, Inst. 2. 7; Cassian, Inst. 3. 3; Cassian, Inst. 7. 27.

(98) Cassian, Inst. 3. 11 [SC 109: 116. 1–2]: die dominica unam tantummodo missam ante prandium celebrari.

(99) Ambr. Ep. 5. 33.

(100) Of course it is equally possible that a later copyist altered whatever term Cassian might have selected to missa.

(101) Cassian, Inst. 3. 5; 3. 6; 3. 8. 2 (twice).

(102) A short list from just the undisputed chapters of Book 3 would include: Cassian, Inst. 3. 1; 3. 2. 3; 3. 3. 1; 3. 7. 1; 3. 9. 2, 3. Additional examples could be adduced from De Institutis 2.

(103) Cassian, Inst. 2. 13 [SC 109: 56].

(104) Cassian, Inst. 2. 15 [SC 109: 56]: Qua lege modestiae post orationum missam unusquisque ad suam cellulam redeat, et cui increpationi subdatur is qui aliter fecerit.

(105) In the course of producing his critical edition, J. C. Guy consulted the oldest extant manuscript of De institutis. This manuscript, a sixth-century palimpsest, is located in the Biblioteca nazionale de Turin, Codex FIV—1 N.16. This partial work contains fragments of De institutis 4. 40–41, 6. 1, and 7. 30–8. 1. Guy noted that while this manuscript did not add anything significant to the later texts, it did support a hypothesis advanced by Petschenig, that the book and chapter headings we find in most manuscripts were added by a later hand. Petschenig had suspected this possibility based on the absence of the headings in his oldest manuscript, Casinensis Rescriptus 295. The absence of these headings in Codex FIV—1 N.16 led Guy to confirm Petschenig's view that the headings were a seventh-century addition for the convenience of later readers (Jean Claude Guy, Jean Cassien: Institutions cénobitiques, SC 109 [Paris: Cerf, 1965], 14).

(106) G. Udney Yule, ‘On Sentence-Lengthasa Statistical CharacteristicofStyleinProse: With Application to Two Cases of Disputed Authorship,’ Biometrika 30 (1938), 377.

(107) William Wake, ‘Sentence-Length Distributions of Greek Authors,’ Journal of the Royal Statistical Society, Series A 120 (1957), 345.

(108) A detailed description of this project may be found in Ivor Francis, ‘An Exposition of a Statistical Approach to the Federalist Dispute,’ in The Computer and Literary Style, edited by J. Leed (Kent, OH: Kent State University Press, 1966), 38–77.

(109) See John F. Burrows, ‘Not Unless You Ask Nicely: The Interpretative Nexus Between Analysis and Information,’ Literary and Linguistic Computing 7 (1992), 91–109, for a complete description of his technique. I am grateful to Professor Burrows for his willingness to read this chapter, in order to validate my conclusions and ensure that I had not misunderstood his technique.

(110) David Holmes, ‘The Evolution of Stylometry in Humanities Scholarship,’ Literary and Linguistic Computing 13 (1998), 114. Holmes also cites a number of studies that have employed the technique.

(111) As a general rule of thumb, the procedure works better with larger samples, as small aberrations are smoothed out over the course of a longer text. In the study of De institutis 1–4, 1,000-word text blocks were used. This length was selected because the questionable text sample (De institutis 3. 4–6, 8) totals 1,017 words. The Burrows Method, as will be demonstrated below, is able to separate samples of known authors, even with blocks this small.

(112) It has been my practice to eliminate words that are context sensitive from consideration when conducting a study of this type. Context-sensitive words would include nouns, most adjectives, and most verbs. These types of words are often related to the subject matter of the text under consideration. If, for instance, one text used the adjective Roman frequently, while another text used the adjective Greek, the differences observed between the two texts would not necessarily imply a different author as much as they would imply a different subject matter. Once the context-related words are removed, what remains are the structural words, those words that occur no matter what subject is treated. Of course it should also be noted that it is quite rare for a context-sensitive word to make the top 50 word list, as these words are usually crowded out by the more common structural words.

(113) The reader who is interested in a fuller discussion of the statistical methodology underlying the Burrows Method is encouraged to consult Burrows, Not Unless You Ask Nicely, 91–109. Multivariate data is transformed into a two-dimensional representation using principal components analysis, and the programmes I have designed to perform this analysis are based on an algorithm found in Ben Bolch and Cliff Huang, Multivariate Statistical Methods for Business and Economics (New Jersey: Prentice-Hall, 1974), 39–40. Another good, introductory description of principal components analysis may be found in Brian Manly, Multivariate Statistical Methods: A Primer (London: Chapman and Hall, 1986), 59–71.

(114) Holmes describes the process this way: ‘Principal components analysis is a standard technique in multivariate statistical data analysis. It aims to transform the observed variable to a new set of variables which are uncorrelated and arranged in decreasing order of importance. These new variables, or components, are linear combinations of the original variables, and it is hoped that the first few variables will account for most of the variation of the original data, thereby reducing the dimensionality of the problem. Typically the data are plotted in the space of the first two components, enabling a two-dimensional graph to portray the configuration of the data in multivariate space. No mathematical assumptions are necessary; the data ‘speaks for itself’. Clusterings of points, each representing a sampled text, are clearly visible, as are outliers which do not conform to any pattern.’ Holmes, Evolution, 113.

(115) Although only three examples have been offered here, further comparisons between Cassian and other Latin authors have been made. In my experience, the Burrows Method has yet to fail to separate works by known different authors. The technique has been confirmed by a number of independent studies (see, for instance, David Holmes and R. Forsyth, ‘The Federalist Revisited: New Directions in Authorship Attribution,’ Literary and Linguistic Computing 10 (1995), 111–27, and Fiona Tweedie, David Holmes, and Thomas Corns, ‘The Provenance of De Doctrina Christiana, Attributed to John Milton,’ Literary and Linguistic Computing 13 (1998), 77–87. Tweedie et al., commenting on the technique states, ‘The “Burrows Technique” as it has come to be called, appears to be a proven and powerful tool in authorship studies’ (Tweedie, Holmes, and Corns, Provenance, 78).

(116) This is an intriguing possibility and clearly further research is needed into the statistical effects of reported speeches in an author's work.

(117) All of the word-frequency averages are given in number of occurrences per 1,000 words. De institutis 3. 4–8 actually contains 1,017 words. As the difference between a standardized value (a rate for 1,000 words) and the displayed value is negligible, the actual word counts for this block will be used for ease of discussion. The standardized values for the five words are: 5.8997 (a), 23.5988 (in), 4.9164 (quoque), 8.84955 (post), and 10.816 (usque).

(118) De institutis 3. 4–8 was not used in calculating the range.

(119) Based on the samples in all of De institutis.

(120) The Z scores for a, in, and quoque are 1.4, 1.27, and 0.87. A Z score is calculated by subtracting the reading from the average and dividing this difference by the standard deviation. 84.13 per cent of all Z scores in a normal distribution should be 1.00 or less; 97.72 per cent of all Z scores should be 2.0 or less; 99.87 per cent of all Z scores will be less than 3.0. Data points that deviate from the mean by a Z score that is greater than 3.0 are thought unlikely to be the result of simple random variation.

(121) David Holmes, ‘A Stylometric Analysis of Mormon Scripture and Related Texts,’ Journal of the Royal Statistical Society, Series A 155 (1992), 118–19.