Towards a Comparative Framework for Research on the Long Cycle in Ethiopic Gospels : Some Preliminary Observations

Setting aside the Gärima Gospels,1 the earliest surviving illuminated manuscripts in Ethiopia have been dated to the late thirteenth and fourteenth centuries.2 Most of these pre­fifteenth­century manuscripts are Gospels,3 with some rare exceptions. However, it is still not known whether this should be taken as an indication that prior to the fifteenth century Gospel books were more often illustrated than other manuscripts or that they were subsequently less likely to be destroyed, or more likely to be preserved, for historical or religious reasons which have yet to be clarified. In these manuscripts, portraits of the Evangelists,4 shown seated or standing, are painted


Introduction
Setting aside the Gärima Gospels, 1 the earliest surviving illuminated manuscripts in Ethiopia have been dated to the late thirteenth and fourteenth centuries. 2Most of these prefifteenthcentury manuscripts are Gospels, 3 with some rare exceptions.However, it is still not known whether this should be taken as an indication that prior to the fifteenth century Gospel books were more often illustrated than other manuscripts or that they were subsequently less likely to be destroyed, or more likely to be preserved, for historical or religious reasons which have yet to be clarified.In these manuscripts, portraits of the Evangelists, 4 shown seated or standing, are painted facing the first gathering of their respective Gospel.The Eusebian prologue, the Canon Tables , and Tempietto 5 -which are sometimes placed after a few folia with introductory textual elements 6 -precede a set of illuminations arranged at the front of the volume. 7hese prefatory illuminations can be categorized into two broad groups on the basis of the number of scenes they include: the short cycle features the Crucifixion, the Holy Women at the Tomb, and the Ascension of Christ; and the long cycle generally includes these latter themes plus a varying number of images showing episodes from the life of Christ, saints, and Old Testament scenes.By the beginning of the fifteenth century, the short cycle was falling into disuse and, by the beginning of the sixteenth century, the tendency to distribute the miniatures from the long cycle throughout the manuscript, between text sections, had become widespread.The aim of this paper is to show that a systematic comparative analysis of the long cycle can improve our limited understanding of how Ethiopian illuminators worked during the early Solomonic period (1270-1527).
While several studies have been published on the short cycle, 8 the long cycle has been less systematically investigated.Indeed, most observations concerning the latter are found in articles which focus on single manuscripts which include the long cycle, 9 and in such studies there has been little room for systematic comparative research.A more comparative approach to the study of Ethiopian miniatures is found in works focusing on a single theme.'The Four Living Creatures of the Apocalypse' by Stanislaw Chojnacki 10 or 'Christ's Entry into Jerusalem in Ethiopia' by Marilyn E. Heldman 11 can be mentioned as examples, as they take into consideration several illuminated Gospel books.However, while such studies further our knowledge of a particular theme, they do less to improve our understanding of the history of the long cycle.Furthermore, they seldom examine the miniatures in a systematic manner (e.g. by translating the captions, describing the manuscripts in which they are found, surveying and comparing all known evidence, etc.).
Indeed, thus far, the only systematic work on a particular iconographic theme is the one published by Ewa BalickaWitakowska, who takes an indepth look at the motif of the Crucifixion without the Crucified in Ethiopic Gospels dating from the late thirteenth to the early fifteenth centuries. 12In this book, the author surveys all the examples of the Crucifixion without the Crucified which were known to her at the time, describing, analysing, and comparing the miniatures.One of the strengths of this publication is that the captions of the analysed miniatures are transliterated and translated 13 -such captions are often neglected, but, as the author recognizes, 14 may provide insight into artistic and scribal practices of the time.Equally valuable is the description of the manuscripts in which the miniatures are found, 15 as codicology is another important but often overlooked aspect in the study of Ethiopian illumination.BalickaWitakowska focuses extensively on image typologies and on literary and iconographic sources, but, understandably one may add, does not tackle the complex issues related to the appearance of this motif in the short and long cycles.Nevertheless, the fact that this Crucifixion type is less consistently found in the long cycle, and that it gradually falls out of the repertoire of Ethiopian illuminators during the fifteenth century, is clearly significant and worthy of further attention.
The above is not to detract from the valuable insights into Ethiopian art provided by the aforementioned studies, but merely to highlight the fact that we are still far from being able to offer a comprehensive overview of the long cycle in Ethiopic Gospel books.This is due to two main reasons.Firstly, because the long cycle includes a large and varying number of images-which display greater thematic and iconographic variety than the short cycle-it poses additional challenges to researchers.Secondly, because the field is still at a relatively early stage of development, the evidence available to researches is sketchy.Suffice it to say here that there are several known examples of Gospel manuscripts featuring the long cycle which have yet to be analysed, published, or described in detail.
In an article on the objectives of Ethiopian art history published in 1955 Jules Leroy asserted that, 'il n'est pas encore possible de donner une histoire de la peinture de l'Église éthiopienne telle qu'elle se présente dans les livres illustrés ou sur les murs des édifices sacrés.' 16 And, in his conclusion, the French scholar observed: Ce dont notre époque a surtout besoin, ce n'est pas d'études multipliées et variées sur tel ou tel point de détail.Ce qu'elle réclame impérieusement, si on ne veut pas s'engager dans des recherches stériles, ce sont des collections aussi riches que possible de reproductions des monuments iconographiques.La tâche est immense dans un pays qui n'a pas encore été prospecté méthodiquement.Si elle paraît humble, elle est cependant d'importance capitale, comme les fondements cachés sur lesquels s'élèvent les murs des plus beaux palais.Tant qu'elle n'aura pas été accomplie, il sera vain de tenter de dessiner l'image de la peinture éthiopienne, en tant que manifestation artistique et historique. 17 In other words, according to Leroy the field needed a large body of reproductions more than studies focusing on a specific topic.This was a reasonable consideration, since it is difficult to draw an accurate history of a particular area of enquiry without surveying all the existing evidence, and this evidence needs to be documented and made available before it can be researched.A few years later, he set out to achieve this goal-together with colleagues Stephen Wright and Otto A. Jäger-by publishing a lavishly illustrated volume on illuminated manuscripts from Ethiopia. 18A similar effort, though more academic in character, was carried out just a few years later by Hammerschmidt and Jäger. 19 Nevertheless, as even wellillustrated publications such as these can only include reproductions of a few of the illuminations from the manuscripts they present, they cannot function as a foundation for a systematic study of illuminated Ethiopic Gospels.
More than fifty years after Leroy's statement much has been done but much remains to be done.The organisation of a series of imaging projects aimed at improving our understanding of the manuscript culture of Ethiopia, such as the Ethiopian Manuscript Microfilm Library (EMML) project launched in 1971, has led to a more methodical approach to documenting and cataloguing Ethiopian manuscripts. 20Moreover, with the advent of the Internet and the invention of digital photography, these documents are becoming more accessible to scholars.In this respect, other projects worthy of mention are the Mäzgäbä Sǝǝlat project-initiated by Michael Gervers and Ewa BalickaWitakowska-and the EthioSPaRe project 21 -led by Denis Nosnitsin.Both projects have a database of images that can be accessed online: 22 the DEEDS database includes a larger number of examples of Ethiopian art than the EthioSPaRe database, but it is less systematically organized and contains less information than the latter which includes reproductions of entire manuscripts as well as a detailed description of their codicological and textual features.As publishing numerous manuscripts in full is not a viable solution, such databases are currently one of the best ways to make complete reproductions of manuscripts kept in hardtoaccess monasteries available for research.
With this increasingly large body of visual evidence being made available, is it possible to say that the foundations for the 'beautiful palace' Leroy hoped to see accomplished have been laid?Almost, for the equally fundamental task of describing and studying this evidence has yet to be complet-18 Leroy et al. 1961. 19 Hammerschmidt andJäger 1968. 20 For a more detailed overview and further bibliography, see Bausi 2004b, 7-11. 21 On this project, see Nosnitsin 2013a. 22There have been also a number of smaller projects, such as the Endangered Archives programme and the Ethiopian Manuscript Imaging Project (EMIP), so far these latter projects have not documented any examples of illuminated Gospels from the early Solomonic period and are therefore less relevant to this paper.While most Ethiopianists are working towards the creation of a shared research environment, somelamentably and to their own discredit-are less open to this idea to the point of even refusing to share their published work online in openaccess journals.
ed. 23 More specifically, with regard to the long cycle, as it is evident that there is still much work to be done, it will be necessary to adopt a stepbystep approach.The field urgently needs studies of each known example of the long cycle which has not yet been the object of a detailed investigation.Ideally, these works should include a description and analysis of: (1) the codicological features and content of the manuscript; (2) the Canon Tables and Eusebian prologue; and (3) a transcription and translation of all the captions.In the future, a scientific analysis of the materials used for painting could yield further valuable data. 24ne aim of these analytical efforts should be to clarify how Ethiopian illuminators operated.There is evidence to suggest that during the early Solomonic period artists and scribes were often ecclesiastics and that patronage played an important role in the commissioning of manuscripts and artworks. 25However, many questions remain to be better answered or explored: how did Ethiopian illuminators operate?How did they connect with the artistic and monastic networks of the country?26Did they travel to work or were the manuscripts taken to them?Was illuminating manuscripts a centralized process, a decentralized process, or a combination of the two?Is it possible to draw a parallel between scribal and artistic practices of the time?Etc.

Comparing the Long Cycle
Admittedly, as the literary sources offer limited information on how Ethiopian artists operated, it is difficult to address these questions. 27However, if the illuminated manuscripts are also examined as a whole and compared they may provide some tentative answers.Indeed, with regard to the long cycle, once the available examples have been catalogued and described, it will also be necessary to establish what their relationship is in terms of their style and iconography.In particular, by comparing the themes used in dif-ferent examples of the long cycle and their iconography, we may gain some insight into how Ethiopian illuminators worked. 28To better illustrate this point it is useful to compare two examples of the long cycle.The first example is wellknown and found in the Kǝbran Gospels (= KG).This manuscript, which was probably decorated towards the end of the reign of Säyfä Arʿad, that is to say towards the end of the third quarter of the fourteenth century, has already been described in detail elsewhere, so it is unnecessary to repeat what has already been said about it here. 29The second cycle is found in a manuscript from a private collection (= PC2).Some remarks about this manuscript have been published elsewhere,30 but, as it has not been fully described and is less known, some additional observations are in order.
Based on its palaeography, PC2, which contains a collection of Marian texts, appears to belong to a period between the seventeenth and eighteenth century. 31Two sets of leaves from earlier periods have been inserted at the beginning of the manuscript.The later of these two sets (fols 5r-14v)which will not be discussed here and can tentatively be ascribed to the second half of the fifteenth century32 -is inserted between the earlier set (fols 2r-4v and fols 15r-18v).On the basis of its style and iconography, the earlier set of miniatures can be approximately dated to a period between the second half of the fourteenth century and the early fifteenth century.While it is likely that this second group of miniatures was originally part of the frontispiece of a Gospel manuscript, this cannot be proven and must be accepted as a working hypothesis.
Unfortunately, I only had a limited amount of time to study and photograph the miniatures in PC2, so I was unable to examine the codicological features of the manuscript in detail.Based on the photographs I have at my disposal, it seems that the single leaves were torn from bifolia and that they were trimmed at the upper and outer sides, probably to fit them in the new manuscript.Captions, damaged and cut to the point of not being entirely intelligible, can be seen above the frame of some of the miniatures.This suggests that the miniatures in which the trimming reaches the very edge of the painted border originally included captions too.As the illuminations of PC2 have been removed from their original manuscript, it will not be possible to compare other features such as the textual evidence and the Canon Tables.In this short paper I will thus focus prevalently on the iconography of KG and PC2.Nevertheless, the two examples of the long cycle discussed here offer fertile ground for comparative investigation.Indeed, setting aside the fact that the artist who painted KG was more accomplished as a painter, it will be seen that the two cycles have much in common while differing in several ways.Table 1 illustrates the subjects that are found in the two cycles and their iconographic relationship.More specifically, Column 1 shows the motifs found in KG, and Column 2 those found in PC2.It is possible that PC2 originally included more miniatures which have now been lost.When a theme is found in both cycles it is highlighted horizontally in grey and, for sake of brevity, it is given an abbreviation which is shown in Column 3. Lastly, Column 4 shows the degree of iconographic affinity, which is measured on a scale from 1 to 6, between miniatures of the same subject in both manuscripts.
The criteria for this latter classification will be clarified in greater detail below, but in general terms they are as follows: (G1) the miniatures have no iconographic affinity and present an entirely different arrangement of elements; we can be confident that they are not related; marginal similarities can be explained in the light of the artistic conventions of the time (e.g.Jesus will always have a halo); (G2) almost no iconographic affinity; the elements are arranged in a predominantly different manner; we have strong ground to suggest that they are not related or very distantly related; (G3) limited similarities in iconography and/or arrangement, the miniatures may be distantly related; (G4) moderate iconographic similarities and/or some similarities in the arrangement; the miniatures are probably related; (G5) numerous iconographic similarities and elements arranged in a similar manner; the miniatures are probably closely related; and (G6) the miniatures have an almost identical arrangement and iconography, to the point where we can be confident that they are closely related.Obviously, the purpose of this scale is to illustrate in a clear manner the strength or weakness of a relationship between two miniatures, but its boundaries should not be taken too rigidly.
The most closely related miniatures in KG and PC2 are those which depict the Arrest of Jesus (= AJ), the Repentance of Peter (= RP), and the Entry into Jerusalem (= EJ).With regard to the AJ (Figs. 1-2),34 the similarities between the two miniatures are as follows: (1) Jesus stands between four armed figures; (2) his left hand is not visible whereas his right hand, palm open, is extended downward; (3) his head is slightly turned to the left; (4) he has a cruciform halo with arms which widen at the outside ends and a rim; (5) he wears a cloak with sleeves and a differently coloured cuff over a tunic decorated with a pattern of vertical parallel lines; (6) the four guards have a turban with one end falling loose behind; (7) the bottom hem of their shirt and skirt is painted in a different colour; (8) three guards turn towards Jesus while the one to the far right turns away; (9) the outermost guards have sticks, the one to Jesus' right holds a sword, while the one to his left holds a rope placed around Jesus' neck;35 (10) their arms are almost bent at a right angle; (11) black geometrical elements with a thin yellow frame appear in the background.There are some minor differences between KG and PC2 in terms of iconography: the guards in KG are depicted in profile; and the object placed around Jesus' neck has two loose ends in PC2 and one in KG.The frames that surround the two miniatures are also quite similar, although the shafts of the columns and the lamp that appear in PC2 are omitted in KG.As noted above, the representations of the RP 36 in KG and PC2 are also closely related (Figs. 3-4): the figures of Mary, John, and Peter are similarly arranged and clothed, and even the folds of the drapery present numerous affinities.They perform exactly the same gestures (i.e.Mary raises her right arm to her chest, John has his arms crossed over his chest, and Peter bends forward lifting both hands to his face).Moreover, in both miniatures there is a cock depicted above Peter's back and we see the same geometric elements which appear also in the background of the AJ.There are no major iconographic differences between the two scenes.Although in KG the col-36 On this subject in Ethiopian illumination, see Heldman 2007, 98-99; Heldman 1972,  181-183; Gnisci 2014, 214-217; and Lepage and Mercier 2011-2012, 132-134.Fig. 2 PC2, the Arrest of Jesus (© Author) umn shafts and lamp, which appear in PC2, are omitted as they are in the AJ.Despite these small differences, the representations of the AJ and the RP in KG and PC2 are so similar that their close relationship seems beyond question.
Fig. 3 KG, the Repentance of Peter (@ Michael Gervers) At the same time, it is evident that both the AJ and the RP in KG and PC2 stem from the same iconographic types transmitted by the Gospel book of Iyäsus Moʾa (= IM), which preserves the earliest known examples of these two subjects in Ethiopian illumination (Fig. 5). 37However, there are some variations between IM and KG/PC2 that need to be outlined.For instance, the cosmic symbols that appear above the miniatures of the AJ and 37 Lake Ḥayq, Ḥayq monastery, c.1280/1281, fol.21v (= RP), fol.22r (= AJ).On the dating and content, see Taddesse Tamrat 1970; Getatchew Haile 1981, 293-301;  Zuurmond 1989, 55-56; Derat 2003, 139-141.On the miniatures, see Heldman 1983;  BalickaWitakowska 1997, 123-124; and BoscTiessé 2010b, with further bibliography.the RP in IM are absent in KG and PC2, in which the miniatures are framed by geometric and architectural motifs which appear in Ethiopian illumination towards the second half of the fourteenth century, and which were possibly derived from CoptoIslamic models. 38Furthermore, the black geometric elements that appear in the background of KG and PC2 are absent in IM.There are other dissimilarities.In the RP in IM there are two cocks instead of one; Mary lifts her hand to her waist rather than to her chest; and the garments have slightly different folds.On the other hand, Peter is depicted with white hair and beard in IM as he is in PC2, so in this respect IM is closer to PC2 than the latter is to KG in which Peter's hair and beard are black.In the AJ the guard to the far left points downward in IM, upward in PC2, and does not point at all in KG, although the position of his arm in this latter miniature resembles PC2.In IM, the arms of the guard to the far right are awkwardly painted-both arms are placed on one side of his body.We find the same mistake in PC2, but not in KG in which the guard's left arm is hidden behind his body and only the hand sticks out.In all three miniatures the guard to Jesus' left holds the rope tied around his neck with two hands.However, in IM and PC2 his right hand passes over the rope whereas in KG it passes more realistically under it.Given that all three manuscripts draw from the same iconographic tradition, it makes sense to attribute these differences to the fact that the artist who painted KG was more accomplished than the illuminators of IM and PC2.At the same time, it is difficult to establish whether, in introducing changes to the iconography of the AJ, the illuminator of KG was being more or less faithful to his model. 39n IM Jesus is presented in full frontal view and points downward with his right index, whereas in KG and PC2 his head is slightly turned to his right and his hand is open.The guard standing to Jesus' right represents an interesting variation to the pattern outlined thus far: in all three miniatures he holds a sword in his right hand, but only in PC2 he holds a second end of the rope tied around Jesus' neck; this is the only instance in which the AJ in KG is closer to IM than it is to PC2.Should we interpret this latter detail in PC2 as a case of 'corruption', to borrow a term from philology, or as a variant deliberately introduced by the artist?Both explanations are plausible and more research will be necessary to resolve the question.Nevertheless, the above analysis shows that the miniatures of the RP and the AJ in KG, PC2, and IM are all related despite some small variations.In the case of these two themes, KG and PC2 are more closely related to each other than they are to IM.At the same time, the two miniatures in PC2 are less distant from IM than KG, though in one respect (i.e. in the detail of the rope in the AJ) KG is closer to IM than PC2.If it is evident that the miniatures of the RP and the AJ in these three manuscripts descend from a common prototype, the origin and date of the model from which they descend remains in question.
The visual evidence suggests that the miniatures of the EJ40 in PC2 and KG are also strongly related (Figs. 6-7), even though they are not as similar as the two groups of miniatures compared above.In both manuscripts, as is conventional in Ethiopian manuscript illumination of the period, 41 the theme of the EJ occupies a twopage spread.The similarities between the left folia are as follows: (1) Jesus rides a donkey and is aligned with the vertical axis of the folium; (2) he rides sidesaddle, holding a book in his left hand and blesses with his right; (3) he is flanked by two lines of figures with halos; (4) these latter figures stand in front of a line of columnlike elements.There are also several differences: (1) in KG there are six figures standing behind Jesus, in PC2 only five; (2) in KG one Apostle holds a crossstaff and another Apostle holds a censer, whereas in PC-2 one Apostle holds a cross and a censer while the Apostle above him points towards Jesus; (3) in KG two children spread their garments before Jesus, whereas in PC2 we see a bearded figure spreading palm branches; (4) all the figures in PC2 turn towards Jesus whereas in KG three figures to the right turn in the opposite direction; (5) in KG the lowermost of these latter figures reads an inscribed page held by a boy, a detail which is absent in PC2. 41 Leroy et al. 1961, 12-13.The similarities and differences between the right folia of the EJ are as follows: (1) Jesus is preceded by two figures (Apostles?) with halos, but in PC2 they turn towards him to greet him whereas in KG they are shown walking ahead of him, although one figure does look backward; (2) both artists seem to have wanted to paint a child with palm branches behind these two figures, but KG features a smallsized figure whereas in PC2 we see the half bust of a youth (another difference which can probably be attributed to the different skill levels exhibited by the two artists); (3) in both manuscripts we see two palm branches floating in the pictorial space and one youth sitting on the shoulders of another and performing similar gestures, in KG the background between these latter two figures and the figures with halos to the far left of the folium is filled with floral elements which are absent in PC2, however there is a fourth youth in this latter manuscript which is absent in KG.Conversely, KG includes a depiction of Zacchaeus while PC2 does not.Yet, the figure of Zacchaeus does not appear to belong to this miniature-he is much smaller than the other figures and is depicted beyond the frame.Therefore, we may assume that this detail was not a property of the main model on which KG and PC2 are based, and that it was added by the illuminator of KG to adhere to an iconographic tradition which was deeply rooted in fourteenthcentury Ethiopian art. 42lso the city of Jerusalem is similarly depicted in KG and PC2.In both manuscripts we see a threecolumned entrance framed by a wall of bricks and topped by an arch.In PC2, however, the bases and capitals of the arches-which are visible in KG-have been omitted, the arch is dislocated, and the vertical lines of the bricks to the right of the entrance have not been painted.Further similarities are the tree to the right of the entrance and the two figures peering out of a polygonal window in the middle register.A series of geometric elements are used to represent the interior of the city, but while these elements are orderly arranged in KG, in PC2 they are chaotically distributed to the point where we can barely distinguish what they represent.Nevertheless, the presence of certain details (e.g. the quadrangular shapes depicted below the window with the two peering figures and the brick structure topped by a triangular roof to its left) in both manuscripts strengthens the impression of their close relationship.
Both miniatures of the EJ are roughly divided into two areas by a series of parallel lines which are surmounted by an interlace pattern flanked by halfcolumns with capitals (pseudoCorinthian in KG and pseudoEgyptian in PC2) in the left folium, and by the interior of the city of Jerusalem in the right folium.A third half column visible in the right folium of KG is absent in PC2, although the element which is visible in the opposite corner of the middle register in this latter manuscript may be a pseudoEgyptian capital.In KG four parallel lines run continuously from fol. 13v to fol.14r, whereas in PC2 the lines-three in fol.3v and five in fol.4r-break in correspondence with the outer frames of the two leaves.In both manuscripts a second partition separates the middle register from the upper areas which are decorated with a trefoil arch filled with a geometric pattern (octagonal in KG and chequered in PC2).While the spandrels of the AJ and the RP in KG and PC2 are filled with a similar type of ornamentation, in the EJ in KG we see a slippalmette motif whereas in PC2 there are small branches with leaves and two facing birds.Moreover, a lamp hangs from the two arches in KG (the one to the left is depicted between the sun and moon), but there are no lamps in PC2.This is in contrast with what happens in the AJ and the RP, in which the lamp is omitted in KG but shown in PC2.
It has been shown above that the trefoil arch, together with other decorative or symbolic elements such as the hanging lamp and ornamented spandrels, was added to iconographic types (the AJ and the RP) used in Ethiopia prior to the appearance of such decorative motifs in manuscript illumination.Now it has been seen that certain details, such as the hanging lamp, do not appear consistently in a particular manuscript.In the light of these observations, I would hazard to say that once these ornamental elements were introduced into Ethiopian manuscript illumination they were used somewhat arbitrarily.At the same time, because the frames and architectural motifs in the themes analysed thus far in KG and PC2 have much in common, as is particularly evident in the case of the EJ, it is reasonable to suggest that Ethiopian illuminators copied the motifs used by other artists introducing their own variations or, at least, not paying too much attention to accuracy.This is hardly surprising.As Zuurmond once observed, with regard to the numerous variant readings in the Ethiopic text of the Gospels, 'it is hardly an exaggeration to say that every Ethiopian scribe in his own way is an "editor."' 43If Ethiopian scribes were not too concerned by the introduction of variations in the text of the Gospels, then it is equally reasonable to suggest that Ethiopian illuminators did not feel the necessity to copy a model with particular accuracy.
It is interesting to observe that, in this instance, there is no evidence to argue that the miniatures of the EJ in KG and PC2 are closely related to the iconographic type transmitted by IM. 44 In other words, the EJ in KG and the EJ in PC2 do not seem to be based on the model on which the EJ in IM depends.Furthermore, while there are a number of miniatures closely related to the iconographic types of the AJ and RP visible in KG, PC2, and IM, to my knowledge, there is no surviving example of the EJ in Ethiopic Gospels from the early Solomonic period that is as closely related to the two miniatures in KG and PC2 as they are related to each other.Only a miniature of the EJ in the Zir Ganela Gospels (= ZG) 45 appears to share more than one or two of the iconographic features visible in KG 46 and 43 Zuurmond 1989, 39. 44 Only the right folium of the EJ in IM survives. 45New York, the Morgan Library & Museum, 1400/01, MS M.828, fols 11v-12r.For a study of the miniatures in this manuscript, Skehan 1954; Heldman 1972; and 'Zir  Ganela', EAe, V (2014), 192a-194a (M.E. Heldman); for reproductions of the EJ, see Skehan 1954, figs. 277-78. 46 Iconographic similarities between KG and ZG have been noted elsewhere, for instance, see Leroy et al. 1961, 9; and BalickaWitakowska 1992, 274-278.PC2, but the ZG miniature also includes a greater number of unrelated motifs.Evidently, as the theme of the EJ is more elaborate than the themes of the AJ and the RP, in that it contains a larger number of elements, it is reasonable to think that the former is more likely to include variations than the latter two.Nevertheless, the above observation can be taken to suggest that certain iconographic types were used more and/or for a longer period of time than others.Unfortunately, in attempting to determine the reasons for this discrepancy, at present it is difficult to go much beyond the evidence offered by the miniatures.For example, it is possible that Ethiopian illuminators had access to a greater number of models for the theme of the EJ, but it is equally possible that certain models were preferred over others for reasons that remain to be explored.It has been shown that the miniatures of the EJ in KG and PC2 are not identical.However, they are sufficiently similar in terms of iconography to suggest that they are closely related.But in what way are they related to each other?And how do we explain the abovedescribed differences between them?With regard to the first question, to focus just on the EJ, the five main possibilities are: (1) PC2 depends on KG; (2) KG depends on PC2; (3) KG and PC2 are based on different but closely related models; (4) KG and PC2 follow the same model; (5) KG and PC2 largely depend on the same model.While none of these options can be ruled out in toto, several observations suggest that the latter possibility is more likely than the others.For instance, with regard to the third possibility, it seems probable that there would be more significant differences between PC2 and KG if they were based on two different models, especially considering the geometric elements inside the city of Jerusalem, which-although differently arranged-are quite similar.Likewise, the inclusion of Zacchaeus in KG suggests that its illuminator had at least another model in mind, and, if accepted, this observation rules out the fourth possibility.
Furthermore, a number of considerations argue against the possibility that PC2 depends on KG or vice versa.Firstly, KG includes several details (i.e. the Apostle reading a passage from a book; Zacchaeus perched on a tree) which are absent in PC2, and it seems unlikely that PC2 would omit both details if KG was its model.At the same time, while the details of the arms of the guard in the AJ can be interpreted as evidence that the illuminator of KG was able or willing to 'correct' an error found in other manuscripts (PC2 and IM), it seems unlikely that he would have been able to make sense of, and copy, the geometrical elements which appear so chaotically beyond the walls of Jerusalem in PC2.
The last consideration does not concern the iconography of the EJ, but that of other themes which are found in both manuscripts.The fact that the two cycles include different subjects makes it unlikely-though not impossible-that PC2 depends on KG or indeed vice versa.However, by comparing the miniatures of the Washing of the Feet (= WF) 47 and the Baptism of Jesus (= BJ) 48 in the two cycles, we can more confidently argue that PC2 cannot depend on KG.In fact, despite the close correlations between the miniatures of the AJ, the RP, and the EJ in PC2 and KG which have been outlined thus far, the miniatures of the BJ and the WF in the two cycles show no signs of a correlation.
In the WF in KG, Jesus is shown standing and in the act of bending forward to wash Peter's feet, whereas the Apostle is shown seated and performing the handtohead gesture which evokes his dialogue with Jesus.This iconography is in stark contrast with the way in which the WF is usually represented in Ethiopian manuscript illumination of the early Solomonic period.In fact, as observed elsewhere, during this period, Christ is conventionally shown seated with his back straight and it is Peter who bends towards him-possibly as an allusion to the liturgy for Maundy Thursday. 49Unlike KG, PC2 is in accordance with this iconographic tradition, which is already attested in IM and is encountered in several other Gospels from this period. 50Furthermore, in PC2 the Apostles are depicted standing and form a square around Jesus, whereas in KG they are seated and lined behind Peter.Therefore, the miniature of the WF in PC2 cannot be dependent on KG, but is most likely indebted to a model which was already circulating in Ethiopia at least by the thirteenth century.Without going into further detail, the same conclusion is also valid for the two miniatures of the BJ. 51If the WF and the BJ in PC2 cannot depend on KG, then it is also unlikely that the EJ in the former cycle depends on that in the latter.In conclusion, with regard to the relationship between the EJ in PC2 and in KG, while we cannot rule out any of the aforementioned possibilities beyond question, the most likely option is that KG and PC2 largely depend on the same model.
So far we have taken into consideration the more divergent cases, namely three themes which are clearly closely related (AJ; RP; EJ) and two which are clearly not (WF; BJ).In such cases, the iconography of the miniatures is so similar or so clearly distinct that there is little difficulty in establishing 47 For a reproduction of the two miniatures, see Gnisci 2015b, figs. 3, 5. 48 For a reproduction of the BJ in PC2, see Mercier 2000b, 55, for the one in KG, see Leroy et al. 1961, pl. 13. 49 Heldman 1972, 175-178;Gnisci 2015b, 253-275. 50 Ibid. 51On this theme, see Chojnacki 1976.whether or not they descend from a common model.It is far more difficult to establish the relationship between those miniatures which are found in both manuscripts and exhibit a moderate degree of iconographic similarity, such as the Ascension (= ASC) and the Crucifixion (= CX).In comparing these latter cases, probably the most challenging task is establishing whether, in what way, and how closely, the miniatures in KG and PC2 are related.More specifically, the key issue is to understand if the two miniatures are the result of a mixing of models, more or less distantly related to an intermediary model, or simply evolved from the same archetype.
The question is a complex one, which no doubt will have to be the subject of further discussion.Taking, for instance, into consideration the CX in KG (Fig. 1), Leroy has observed that it stands apart from most other Ethiopian miniatures of the period, in that 'it conforms in every detail to the Byzantine pattern.' 52 Also for Heldman several miniatures in KG 'follow almost contemporary Byzantine models.' 53 In her opinion, the miniatures in KG 'may be divided into groups which reflect different models.' 54One subdivision is represented by the miniatures of the AJ, the RP, the ENT, and the RJ, which follow 'a Passion cycle that often appears in Ethiopian illuminated Gospel book from the early years of the fourteenth century.' 55 It is worth emphasising that this subgroup includes all the miniatures (apart from the EJ) which are most closely related to the miniatures in PC2.
Heldman's second subdivision includes the following 'Byzantinizing' 56 miniatures: the four Evangelists; the Annunciation; the Nativity; the Presentation in the Temple; the Adoration of the Magi; the BJ; the Transfiguration; the Descent from the Cross; the WF; and the Anastasis.As observed elsewhere, 57 most of the miniatures in this latter group offer either the earliest known example of a theme-such is the case of the Deposition and the Anastasis-or the earliest known example of a particular iconographic type-such is the case of the WF, the Presentation in the Temple, and the Nativity-in Ethiopian illumination.Significantly, Heldman does not comment on Leroy's analysis of the CX and does not focus on the miniatures of the EJ, the CX and the ASC in KG.With regard to the ASC and the CX this may be because, as noted above, these two miniatures are more 52 Leroy et al. 1961, 11. 53 Heldman 1979b, 359. 54 Ibid. 55Ibid. 56Ibid. 57Ibid.See also, BalickaWitakowska 1989, 15; Chojnacki 2009, 24; Gnisci 2015c, 561-568.problematic than others, in that they appear to be only partly indebted to earlier models.This can be illustrated, for example, by comparing the miniature of the CX in KG with that in PC2, and in doing so, it becomes soon evident that Leroy's conviction that KG conforms to a 'Byzantine pattern' warrants further scrutiny.In PC2 the cross is empty, surmounted by the Lamb of God, and flanked by the two thieves with the spear bearer and the sponge bearer.In other words, PC2 adopts the iconographic type for the CX that was predominant in fourteenthcentury Ethiopia. 58If we don't consider the addition of decorative elements, that is to say the columns and the trefoil arch, the CX in PC2 matches the type seen in IM and in most other Ethiopic Gospels from the fourteenth century.In fact, the only difference between the CX in IM and in PC2 are the shapes of the crosses and the fact that the hemisphere with the sun, moon and the mountains 59 which are visible in the background of IM have been lowered in PC2 to allow the introduction of the trefoil arch in the upper register.
The CX in KG presents us with a more complex picture.On the one hand, we find several of the elements that are also visible in PC2, namely the two thieves, the sun and moon, the cross, the spear bearer and the sponge bearer, whose poses recall those seen in PC2.These elements are common in Ethiopian depictions of the period but are not common in those nearcoeval Byzantine models which, according to Heldman, influenced the second subgroup of miniatures listed above.On the other hand, in KG we also see a series of elements which are absent in other miniatures of the CX in fourteenthcentury Ethiopic Gospel books, namely the two angels who cover their mouths 60 and the body of Jesus on the cross. 61We find a similar combination of old and new elements also in the ASC in KG (Fig. 8).The upper register, which features Christ in a mandorla supported by the tetramorph, recalls numerous details of the ASC in PC2 (Fig. 9) and in a num- 58 For a more detailed discussion and several examples, see BalickaWitakowska 1997. 59On the identification of this detail, see Heldman 1979a, 110. 60  ber of other fourteenthcentury miniatures of this subject. 62However, the Apostles who witness the Ascension in the lower register of KG have natural poses which exhibit little or no relationship with the hieratic rigidity that is visible in the ASC in PC2 and in most examples of this subject in Ethiopic Gospels of the fourteenth century. 62For instance, see the disposition of the figures in the upper register of the ASC in Baltimore, The Walters Art Museum, first half of the fourteenth century (?), acc.no.W. 836, fol.7v.In light of these observations, it is natural to ask whether the CX and ASC in KG are the result of a mixing of models.The question will remain open to further debate and my goal here is not to identify the iconographic sources of KG.Nevertheless, the visual evidence strongly suggests that the two miniatures, which include new or uncommon iconographic elements for the period, are partly indebted to models that were already circulating in Ethiopia between the late thirteenth and early fourteenth centuries.The fact that the illuminator was willing to adapt or modify the models he had at his disposal to meet his aims or those of his patron reinforces this possibility.The artist's predisposition to adapting his models is attested by a number of other miniatures.It is illustrated, for instance, by the inclusion of Zacchaeus in the EJ, which has been discussed above, and by the introduction of an Ethiopian altar in the Presentation in the Temple, which was first noted by Heldman. 63 Another clearly visible feature in the cycle in KG is a propensity to present scenes featuring Jesus in a more naturalistic, or less 'respectful', manner.This can be seen in several of the miniatures, but suffice it to mention the Apostles who turn away from Jesus in the EJ and the fact that Christ bends forward in the WF.This is in stark contrast with the 'cultural or religious' bias which characterizes the work of most Ethiopian illuminators during the fourteenth century.In fact, as noted elsewhere, 64 especially between the late thirteenth and early fifteenth centuries, Ethiopian illuminators exhibit a marked tendency to underscore the fact that Christ is God by adopting uncommon iconographic solutions.This can also be seen in the miniatures of PC2, as shown by the fact that all the figures in the EJ turn towards Jesus or by the disposition of figures in the WF.It is difficult to determine at this stage whether the choices of the KG illuminator were influenced by artistic models or motivated by theological reasons.
If we take into consideration everything that has been said so far about the iconography of the cycles in KG and PC2, it becomes possible to draw a few preliminary conclusions.More specifically, while it has been observed that the stylistic features of the two cycles suggest that they cannot be too chronologically distant from each other, it is evident that the cycle in PC2 is more indebted to models that were circulating in Ethiopia by the end of the thirteenth century than the cycle in KG.This can be demonstrated, to an extent, by comparing the themes that are found in KG, PC2, and IM and establishing how closely related they are to each other.As illustrated in the table below, PC2 is almost equally related in terms of iconography to both KG and IM, whereas KG is considerably more distant from IM than PC2.This could suggest that the cycle in PC2 is slightly earlier than the one in KG, but it is equally possible that they are roughly coeval and that the illuminator of KG was simply more innovative, or, at least, had access to more recent models.

Conclusion
In conclusion, and emphasising once again the preliminary character of these observations, a comparison between the cycles in KG and PC2 can help us improve our understanding of how Ethiopian illuminators worked.Firstly, it is evident that certain iconography types-such as the RP and the AJ-enjoyed a longer and more widespread circulation than others-such as the EJ.At this stage, it is only possible to speculate on the reasons for this discrepancy, but a number of possibilities come to mind.For instance, it is possible that Ethiopian illuminators had access to more models for a certain subject, but it is also possible that in some cases certain models were preferred over others for reasons that remain to be clarified.
Secondly, it has been demonstrated that both illuminators drew on a variety of sources to create the prefatory cycles in PC2 and KG.This is particularly evident in the case of KG, which appears to draw from at least four different sources: (1) a first group of miniatures (AJ; RP; ENT; RJ) follow models which are already attested in IM; (2) a second group of miniatures, which have been described as 'Byzantinizing', adopt iconographic formulas that are unattested in earlier Ethiopic Gospels (e.g.WF; BJ; the Nativity; the Presentation in the Temple; etc.); (3) a third group of miniatures (CX; ASC) seem to be the result of a mixing of models; (4) and, finally, one miniature (EJ) seems to be based on a model followed closely only by PC2.It seems unlikely that the model for this latter miniature was part of the source or sources used by the illuminator of KG for painting the second group of illuminations because of its close relationship with the EJ in PC2-which exhibits no relationship with this second group of miniatures.These observations also indicate that the process which led to the updating and revision of certain iconographic types was a gradual one.
Thirdly, by comparing the differences which characterize themes such as the AJ in KG and PC2 we can determine that Ethiopian artists were not compelled to be too faithful to a particular model.For instance, the manner in which the illuminator of KG paints the arms of the guard to the far right in the AJ attests a willingness to 'correct' certain details-though we cannot entirely rule out the possibility that by not following PC2 and IM the illuminator of KG was being more accurate to the prototype as this has yet to be reconstructed and identified.In any case, given that the miniatures in all three manuscripts are related to the same model, it is unquestionable that either KG or PC2 and IM are being less faithful to the prototype.We may also come across variants (e.g. the rope around Jesus' neck) that are not always easily explained.
Lastly, by analysing and comparing the miniatures we may gain further insight into the artistic practices of the time.For instance, with regard to the EJ, the inclusion of Zacchaeus in KG, suggests that an artist could draw from multiple sources to paint a scene.Other differences can probably be attributed to the different degree of accomplishment attained by the two illuminators.More specifically, it is tempting to view the use of halfbust figures to represent children and the chaotic distribution of the elements which constitute the city in PC2 as evidence of how quickly corruptions could appear in the copying process.The different number of figures in the two miniatures could also constitute a case of corruption.Lastly, the varying attitudes of the Apostles (those in PC2 all turn towards Christ whereas in KG four figures turn in the opposite direction) can be attributed to the different 'stylistic choices' of the two illuminators.This is about as far as we can go based on the evidence analysed so far.However, by taking into consideration the other themes found in KG and PC2 as well as other illuminated Gospels it should be possible to push this analysis a little further.Moreover, by adopting a more systematic approach to the study of illuminated Ethiopic Gospels and taking into consideration all the evidence offered by the manuscripts we may attempt to identify correlations between artistic and scribal practices.To call for a more holistic approach to the study of Ethiopian manuscript illumination-one that also takes into consideration the codicological, philological, and palaeographic evidence-is not to encourage the development of a field of 'manuscriptology', as some might fear. 65Rather, on the one hand, it is to emphasize the importance of going beyond the visual evidence to broaden the scope of the art historical analysis and trace the development of illumination in Ethiopia as an integral part of a more comprehensive history of manuscript produc-tion and transmission.And, on the other hand, it is to encourage collaborative research.66Indeed, given the vast amount of manuscripts that have yet to be catalogued and analysed, a teambased approach seems the best way forward to offer a framework, which needs to be both quantitative and qualitative, for investigating Ethiopian manuscripts.

Fig. 1
Fig. 1 KG, the Arrest of Jesus and the Crucifixion (© Michael Gervers)

Fig. 5
Fig. 5 IM, the Repentance of Peter and the Arrest of Jesus (@ Michael Gervers)

Table 1 Comparison between the miniature cycles in KG and PC2
As this detail is found in both Eastern and Western icons, it is difficult to pinpoint what model inspired the illuminator of KG.For instance, to mention a context close to Ethiopia, icons featuring this detail are found in Saint Catherine's Monastery, Mt.Sinai, see Weitzman 1978, pls.26, 38, for reproductions. 61While KG represents the earliest witness to the diffusion of this motif in illuminated Gospels, in the context of Ethiopian art, the living Christ on the cross appears on a