Byways in British Archaeology

The first comment on such a catalogue concerns the interpretation of the figures. Obviously, it is the height of the church relatively to the village or valley which is of importance, not the actual elevation above the sea. For example, although the Sussex village of Bishopstone lies in a chalk coombe, the church (Fig. 23) is built on a hillock which overlooks the dwellings of the inhabitants. Again, with respect to villages situated on the plain, the church frequently occupies the highest ground. Various causes may have been at work to influence the selection of such spots. Here, there were difficulties with the feudal lord (p. 17 supra); there, the building was placed centrally so as to serve numerous outlying townships; yonder, the church stands prominently on the hill-top because a hamlet had arisen there before the church was thought of.

When all these deductions have been made, there remain very many churches unaccounted for. These have been raised, after prolonged labour and at great expense, on some inconvenient hill, up which the worshippers must struggle breathlessly Sunday by Sunday. Standing by one of these churches, the observer is mastered by the conviction that he is placed on an ancient vantage ground. Range of outlook, and effective strength of position, must have been the fundamental ideas in the minds of those who chose the site. Throw aside this conception, and all is contrariety. The modern architect, unaided by the student of folk-lore and folk-custom, is entirely at a loss when asked to explain such apparent folly on the part of the designers.

We have before noticed (p. 17 supra) how the annals of folk-lore are crowded with stories respecting these hill-top churches. The tales vary a little in detail, but the main theme is the same. Devils, or witches, or fairies, were in league against men, and so soon as the builders began their work, unseen hands removed the stones by night and laid them elsewhere, the church being finally set up only after a severe and lengthened contest. A variant of the superstition is met with in Scotland and Brittany, where churches are reported to have been reared in a single night by the fairies[269]. Just noting, as we pass, that the legends do not attempt to tell why men were so anxious to have their churches in lofty positions, let us glance at some of these bewitched buildings.

One of the examples best known to the writer is that of Churchdown, in Gloucestershire. The church stands on a Liassic hill of pyramidal outline, 500 feet above the sea-level. To climb the steep slope from the village below is an arduous task. The devil-theory still thrives vigorously in the neighbourhood, and the visitor will soon hear the account of the nightly removal of the stones by the spirit of evil. It chances that the local pronunciation of Churchdown is “Chōsen,” and this has suggested to the otherwise serious countryfolk a weak pun. Not only, they say, was the spot deliberately “chosen,” but the “chosen people” of the Prayer-Book response refers specifically to the favoured inhabitants of the village. The traveller cannot escape—

Breedon church, in Leicestershire, is on a hill which similarly overlooks the village clustered at the foot. For pedestrians, the ascent has been made easier by cutting steps in the pathway, but carriages have to take a circuitous route to reach the church[270]. Other churches concerning which a contest was waged between Christian masons and the powers of darkness are St Chad’s, Rochdale, and Capel Garmon, in Denbighshire. In the second of these cases, the church was to have been erected on a hill, contiguous to an ancient spring, but the stones were repeatedly carried to a lower position[271]. This detail respecting the spring is not without significance, as was shown in the last chapter. Another illuminative Welsh example comes from Llanllechid, near Bangor. Here the fairies, or spirits, bore away the stones from a field called Caer Capel, and actually selected a situation—so the legend runs—more convenient than the first[272]. Both Caer Capel and Llanllechid are suggestive names, the latter gives a hint of a church built on the site of megaliths. At Wendover, in Buckinghamshire, the builders were thwarted by malevolent witches: at Hanchurch and Walsall, in Staffordshire, the trouble was caused by mischievous fairies. Each locality has its own bit of colour tinging the testimony, but to rehearse all the variations would be tedious. The number of these solitary hill-top churches was once probably much greater than at present.

In some instances the buildings have been demolished; in others, modern dwellings have sprung up around the church and have masked its isolated position. Indeed, this shifting of the population is a factor which must never be left out of sight.

How shall we explain the superstitions attached to these churches? One fact is manifest—the selection of site was, in the majority of cases, freely made, in the sense that other spots were available. On a broader interpretation, based on retrospect, we perceive that there was little choice, since there existed an ever-present necessity for an asylum in times of stress and danger, not to mention minor reasons. We have seen that the traditions do not assign a cause for the persistence,—nay, obstinacy, of the church-builders. Perhaps we should except those churches which were erected in particular spots as the result of the vision of some saint, or of warning cries and mysterious voices heard by the masons. On the whole, however, we can interpret the stories only by the aid of folklore and anthropology. Modern research has shown that, in the earlier stages of human life in Britain, the communities dwelt mainly on the downlands and elevated moorlands[273]. Only when the historical period was well advanced, namely, during the late Roman and Saxon occupations, were the thickly wooded valleys and marshy plains settled and cultivated. The evidence, though largely inferential, is both good and abundant, and need not be recapitulated here. All that is essential, for the moment, is to remember that descendants of the earlier races probably lived on, isolated in sparse communities, wherever the conditions were favourable. This is the legitimate conclusion to be drawn, for example, from the classical excavations made by Pitt-Rivers in the chalk downlands of Cranborne Chase.

Anthropological tests afford similar support in other regions, such as the Yorkshire Wolds and the Derbyshire Moors. The remnants of the older British races, modified somewhat by intermarriage, doubtless held aloof on their bleak hills and tablelands with considerable stubbornness. The Teutonic farmers of the vales would not at first greatly interfere with the hill-top folk. A time came, however, when fresh-comers began seriously to re-invade the more elevated spots, and to form settlements. This does not seem to have been done systematically until after the Norman period. But it is clear that any impinging of new races upon the older hill-top communities, however partial in its character, would give birth to exaggeration and myth. A fair corollary may be stated: the traditions which tell of the efforts of builders having been baulked or defeated by demons, witches, or evil spirits, are echoes of the time when older races still lingered sparingly on the hill-top settlements. On the other hand, the legends which speak of dreams and visions seem to represent a purely religious development belonging to the Christian period.

The erection of a church on soil hitherto relegated, by silent consent, to primitive tribal folk, or, failing these, to the wild beasts, was a novel and startling event. We need not assume that the primary purpose of the builders was to Christianize those remote peoples, though that may have been a subsidiary motive. It is more likely that the church was intended for the churchmen who built it. But to enter an area previously left intact would provoke keen opposition. Further, it is now a commonplace that our stories of witches and wizards, fairies and demons, are in part, at least, derived from folk of the Neolithic and Bronze periods, or from their near descendants. To the conquered people was imputed magical power, while to the victors, their weapons, and their appliances, the beaten folk paid the respect due to fear. Briefly, then, the superstitions which we are considering seem to speak of the projection of one race over territory long occupied by the survivors of much earlier races, and again, of the intrusion of social habits and religious customs among folk whose beliefs and modes of life were widely different from those of the invaders.

Still the question is unanswered, Why were hill-tops chosen as sites of churches? Primarily, of course, the churches were built for worship, but the answer to the question seems to be, that, in very many cases, the building was intended for temporary refuge and defence in those unsettled centuries which lasted down to the Wars of the Roses (cf. p. 57 supra, concerning earthworks). To establish this theory, we must consider a moderate number of instances extracted from a lengthy list. And a clearer understanding will be promoted if our minds are chiefly fixed, for the present, upon church towers alone, because these are the portions of the buildings which best illustrate the theory.

As might be anticipated, the specimens of towers which furnish the most apposite illustrations are found in the Border Counties, and in the Welsh Marches, or, on the other hand, in those districts which have few natural defences. Beginning with Cumberland, we have the embattled tower of Great Salkeld church, guarded by a massive door, plated with iron, and fitted on the inside with stout iron bars. Under the aisle of the church are chambers, which are believed to have been used as dungeons[274]. The tower at Burgh-on-the-Sands, near Carlisle, seems to have served as a fortress or pele-tower[275]. (Etymologically pele or peel = the stockaded enclosure or fort.) Normally, the pele or “peel” towers were special structures, and were erected along the border irrespective of the position of churches, as at Corbridge, in Northumberland[276]. Nevertheless, there is little doubt that strong church towers were used as substitutes for peles in times of necessity. Thus, the twelfth century tower of Edlingham church, in Northumberland, is one of several that served this purpose; while Merrington church, with its tall steeple, crowning an eminence, actually stood a siege. The tower of Bedale church, near Richmond in Yorkshire, seems to have been constructed purposely for defence. The narrow staircase has, in fact, a portcullis groove, which was accidentally revealed after the steeple had been struck by lightning. The portcullis screened the tower from the body of the church[277]. Melsonby tower, also in Yorkshire, has been described as “a Norman keep in miniature[278].”

The round steeple of the Early Decorated church of Roos, in Holderness, is supposed to have been a post for watch-and-ward. It stands on high ground, and contains, near the top, a chamber which is reached by a spiral staircase of stone[279]. We know that the surrounding district was terribly ravaged by the Danes, and it is interesting to note that, in 1836, there was discovered, in a ditch at Roos, a rude model of a boat with a warrior crew. This curious object is considered to be of Scandinavian origin, and to date back to the invasion of the Danes and vikings[280]. If it be urged that the church fabric belongs to a period long subsequent to the raids of the Northmen, the comment is both fair and relevant. Two replies might be made: first, analogy indicates that churches of this kind most likely had predecessors on the same site, and secondly, defensive structures continued to be built long after the events which called them into existence passed away.

From Yorkshire we pass into North-East Lincolnshire. Here we meet with a remarkable series of Saxon towers, and here the Danish question recurs. Popular tradition, vigorous even “yn tyme of mynde,” as Leland quaintly expresses it, declared that the present towers were built as refuges from the Danes. The idea was unhesitatingly accepted by the earlier school of antiquaries, and was supported by the older architects. We now know, as the result of comparative study of architectural styles, that the theory is barely tenable. The three best-known towers of the group under consideration are those of Waith, Scartho, and Holton-le-Clay, and they are all placed by Professor Baldwin Brown in his Saxon period C[3]. In other words they are believed to belong to the time of Edward the Confessor, when there was a revival of church-building[281]. As a further safeguard against ante-dating these churches, we may notice Mr Francis Bond’s dictum that we have no Anglo-Saxon tower which is earlier than the end of the ninth century[282]. Of course, neither of these verdicts affects the question of the existence of previous towers on the present foundations. For instance,

Fig. 24. Scartho church, Lincolnshire, from the South. The Saxon tower differs from the ordinary defensive type in having an original doorway at the base, seen at the West end. The doorway facing the spectator is a later insertion (Early English) and the parapet belongs to the Perpendicular period. There is a Saxon belfry window, divided by a mid-wall shaft. The walls separating the tower from the nave are three feet in thickness.

according to writers of the middle of the last century, and indeed, much later, the tower of Scartho church (Fig. 24) was said to rest on large blocks of stone which showed traces of fire, and these blocks were thought to be relics of an older fabric which was burnt by the Danes. I have visited this church several times, and, while believing that the present tower may not be the original, I am not convinced that there are any fire-marked boulders to be seen. True, the masses of stone referred to by the topographers may now be quite concealed under the raised turf, but it is more likely that red lumps of ferruginous grit, which are plainly visible, have misled the antiquaries. The tower walls certainly contain a miscellany of geological specimens,—limestones, flints, sandstones, and rusty-coloured grits—and superficially the stains suggest scars caused by fire. One must, however, take account of all the circumstances that influenced church-building in a locality which, like that under consideration, is destitute of building stone. Some two or three miles distant is the church of Clee, the tower of which, except in the uppermost stage, bears a close likeness to that of Scartho. (The parapet, in each church, is a later addition.) The materials employed in Clee tower are mainly large beach pebbles, and ice-worn “cobbles,” which had been washed out of the now-vanished cliff of Boulder Clay at Cleethorpes, and which must have been assiduously collected by the perplexed and needy masons.

Professor Brown declares that there is nothing about the Lincolnshire towers themselves to indicate a defensive character, though they are situated in a region which was exposed to Danish visitations. These eleventh century towers, however, he says, represent a style which was evolved at a somewhat earlier time, when the Danes were actually hostile. At most, he will admit only “a certain general likelihood” that these particular towers were used for defence[283], but he notes the general opinion that the occurrence of such towers in districts like Lincolnshire is “hardly fortuitous[284].” The palmary fact remains, that the Lincolnshire towers were, relatively to the exposed and unprotected tract of marshland on which they were built, real fortresses. There are no natural defences in the vicinity of the churches, and ordinary dwellings must have been of an insubstantial kind. Moreover, unless the idea of earlier buildings on the present sites be unreservedly abandoned, there still remains the great probability of continuity of general pattern. Long after the land ceased to be harried by pirates, the countryfolk would anticipate, and provide against, future onslaughts. It is no bar to this argument to oppose the fact that, as we later folk now know, those attacks were not made. We shall, indeed, find that defensive towers continued to be popular for several centuries after the Norman Conquest. We may say then, that the Lincolnshire type of Saxon tower was semi-defensive, though the existing specimens are of post-Danish age.

Leaving Lincolnshire, we notice a Bedfordshire example, that of Clapham. Here the church has a massive Saxon tower, which possesses an exterior door at a height of twenty feet from the ground. There is no window opening in the lower portion. It is not an insignificant fact that this tower overlooked a ford across the Ouse, which was approached by an ancient road[285]. The striking position of the door at Clapham has many parallels. The old tower of Swanage parish church, Dorset, had no door lower than the second story, and access was probably gained by ladders, which could be afterwards drawn up if required. At Norton, Derbyshire, the doorway of the tower, which is supposed to have been erected about A.D. 1300, is six feet from the ground, and was formerly reached by an external staircase of stone[286]. Occasionally, the external door was absent altogether. This was originally the case at Rugby church, the tower of which could be entered only through the nave[287]. This tower, which dates from the fourteenth century, is of square outline, and is devoid of buttresses. Its loftiness is conspicuous, for it reaches a height of 63 feet. The windows are narrow, and the lowest is twelve feet from the ground. Consider, again, the little church of Swindon, near Cheltenham, with which the writer was once very familiar. The tower is of hexagonal shape, and the walls are thick and substantial. The only windows, a quarter of a century ago, were mere loopholes, splayed without and within, and placed near the top of the structure.

Fig. 25. Oystermouth church, Glamorganshire, from the South-West. The tower is less lofty than is usual with the churches of Gower, but it has several “defensive” features. It batters from the base, and has the characteristic battlement. There are no buttresses, but a flat staircase-turret will be noticed. Except the debased West window, the openings are all of very insignificant size.

The defensive tower, however, is perhaps seen at its best in South Wales. In the district of Gower, in Glamorganshire, twelve out of sixteen towers are believed to have been erected as much for defence as for beauty; each is a stronghold as well as a campanile[288]. Generally these towers have no buttresses, and frequently, as at Oystermouth (Fig. 25), they “batter,” or slope a little from the base upwards. This battering is also met with in the castles of South Wales. Again, the towers have an overhanging embattled parapet, supported on corbels. Where exterior doors are now found, they are usually later insertions. The original doors, where they exist, resemble those of keeps in being situated some height from the ground. The tower, too, opens into the nave as often by a mere doorway as by a belfry arch. Instead of belfry windows, there are insignificant slits, and the rest of the wall is quite blank, and destitute of architectural or decorative features. In short, as Freeman expressed it, the essential military character of the towers is stamped on every stone[289].

The churches of South Pembrokeshire are not less remarkable. The study of these buildings is rendered more attractive because the area in which they are found corresponds roughly with the territory occupied by the non-Welsh folk of “Little England beyond Wales.” This fact may indicate a period of severe struggles between the English and Norman intruders and the earlier settlers, but not necessarily; since the Gower churches, at least, would have to be explained somewhat differently. The raids of the Norsemen upon the Pembrokeshire coast seem to have come to an end with the eleventh century; hence dread of inland foes seems to have caused the erection of these strong towers. Professor Freeman, who carefully investigated the Pembrokeshire towers, and who wrote a painstaking account of their architectural details, observed, “Every tower is a fortress, designed to hold out as long as Zaragoza or Sebastopol[290].” Anyone who has examined these churches, will, after making some deduction for the literary form of expression, be prepared to endorse this opinion. Take, for instance, the church of Gumfreston, near Tenby (Fig. 26), already known to the reader through its medicinal spring (p. 95 supra). This sequestered church is built on the inner slopes of that high ridge which, on the North, overhangs the lovely vale of St Florence. Almost enclosed by trees, with whose foliage the ivy-wreathed tower quietly harmonizes, the tall fabric is approached unwittingly, so that, when the visitor reaches the gate of the churchyard, the

picture calls forth a sudden exclamation of pleasure. A second and more critical glance reveals the chief features of the tower. Its height, 65 feet, is especially noticeable. Like other towers which we have met, it tapers a little, and it is crowned with a battlement. Each of its five stories was evidently devoted to a particular use. Freeman has noted them: a ringers’ chamber below, followed by stories which had windows looking towards the North and East, then a room fitted up as a dovecot, and lastly the belfry itself[291]. The date of the Gumfreston tower is fixed approximately at A.D. 1300. Of the Pembrokeshire series as a whole, Freeman says that they were all built within “castle times,” and that they belong to “all manner of centuries from the first to the last Harry[292].” He further considers that he could discern preparations for habitation in “some of the vaulted apartments.” The simple character of the Pembroke towers formerly led many writers to assign them to a pre-Norman period, but this opinion is now discredited. On the other hand, the evidence shows that they are not, as a class, late Tudor buildings, as extremists in the contrary direction have suggested.

A subtle objection is made that the towers may be simply imitations of castellated architecture, and that the resemblance, being incidental and unintentional, has no recondite significance. There is some degree of force in the contention, because there must have been interaction of influence. Thus, not only in semi-secular buildings, like the abbot’s barn of Bradford-on-Avon (Figs. 43, 44, pp. 171, 173 infra), but also in castles and fortified mansions, like that of Nunney (Fig. 27), do we meet with architectural features which are usually associated with religious structures. Such details, as a rule, are best seen in doors and windows, and they are in harmony with the Gothic styles of their respective periods. Side by side with these ornamental features, primitive arrangements for defence are retained, even in the Tudor country mansions. In spite of this cross-influence it may be observed that the defensive steeples occur just in those places where defence

Fig. 27. Portion of Nunney Castle, Somerset, a fortified manor-house, built A.D. 1373-1400. The corner towers probably served as peles. In the windows (Decorated style) and in other details, there is evidence of the relationship between religious and secular architecture.

would be required. Often, as at Gumfreston, the church was the only building strong enough to give adequate security to human beings and to portable property. Castles do not exist everywhere, neither, indeed, do these sturdy towers; but the one group seems to be largely complementary to the other. Exception has also been taken to the evidential value of the narrow window openings. Mr Allcroft, while expressing agreement with the theory that ecclesiastical towers were frequently employed for protection, deprecates an appeal to these narrow slits as affording satisfactory testimony[293]. He contends that the openings were made strait and were provided with an inner splay to minimize draughts. That they would be moderately efficient for this purpose may be conceded. Not only the ordinary current of air, but the high gale, had to be foreseen and provided against. Glazing the apertures was out of the question, in the early times when glass was a costly article, especially in remote counties. But there is another mode of approaching the question. What was the purpose of the well-known “oillets” in buildings purely defensive? Even in primitive warfare no tower was impregnable if it presented wide openings to the foe. Arrows and slingstones would, under certain conditions, prove more swift and deadly agents than fire itself. We may grant that the round-headed twin aperture, with its single plain baluster-shaft in the middle, so familiar in the late Saxon towers of England, was not avowedly of a defensive character.

That admission does not exclude a belief in the defensive value of the narrow window. During the Perpendicular period a tiny battlemented parapet was frequently used for the ornamentation of a capital or transom, but such a practice does not negative the original use of the battlement in fortification. Further it might be reasonably argued that the strait window-opening was copied from secular strongholds because it originally served one purpose in all cases. The point is not of prime importance, but it is worthy of note that the narrow loophole continued to prevail after glass had come into general use for church windows, and after bells had become common. The price of glass was, therefore, not the deterrent. Again, the bell-ringer would perhaps have actually welcomed a wider opening. Whatever may have been the motive, we find that, from the Early English period onward, the tiers of spire lights known as dormer window’s or lucarnes, though mainly decorative in character, continued to be made very narrow. On the other hand, these dormers were so arranged that one series was placed, with respect to the group above or below, on alternate sides of the spire. One fact remains; in towers which were designed to withstand attack, the retention, if not the initiation, of the narrow type of opening, suggests motives of strategy.

Before quitting the English towers it is well to note that the strength of many of them was tested during the Civil War. In Devonshire alone, a number of instances can be brought forward. The Parliamentary troopers turned the towers of Powderham and Ottery St Mary into temporary fortresses, while the Royalists annexed those of Tiverton, St Budeaux, and Townstall[294]. In Wiltshire and Yorkshire, again, several churches were used as shelter for men and horses. Thorold Rogers asserts that the Royalists garrisoned the parish church of a Hampshire town, the name of which, however, he does not give, and, thus protected, withstood a siege and cannonade[295]. The reader will doubtless recall many instances in which tradition tells of church walls battered by shot and shell. Making due allowance for exaggeration, it must be admitted that folk-memory has a sound basis in fact, and some, at least, of the damaged buildings may be genuine illustrations of the effects of assault by artillery.

It is all but impossible to bid farewell to defensive steeples without some reference to the famous round towers of Ireland. The disinclination to ignore the subject is increased by the knowledge that these structures afford collateral proof of the defence theory. For the best instructed modern opinion pronounces them to be fortresses as well as belfries. These towers, which have been the cause of an abiding controversy, are typically tall and slender, and are surmounted by conical caps. Their height ranges roughly from about 60 to 125 feet. The number of examples still remaining is given as 76, and they are found exclusively in association with some early church, or with some ecclesiastical settlement. The usual position is near the North-West door of the church[296]. This fact alone indicates their connection with early Christianity. A typical round tower, somewhat restored, is that of Devenish, situated on an islet of the same name in Lower Lough Erne, county Fermanagh (Fig. 28). In the vicinity is the ruined priory of St Molaise, as well as several other ecclesiastical remains.

Fig. 28. Round tower of Devenish, on Devenish Isle, Lower Lough Erne, near Enniskillen. Height, 84 feet; doorway, 9 feet from the ground; walls, 4 feet thick. The windows approximately face the Cardinal Points. A string course will be seen around the base of the conical cap.

A gradual advance can be traced in the architectural style of the round towers. The stages of their development were parallel to, and probably contemporaneous with, those of the churches and oratories[297]. Three periods of building have been distinguished. It is not until the beginning of the tenth century, says Miss Eleanor Hull, that we find the towers mentioned, and none can be dated earlier with certainty. They are later in date than the churches built by St Columba at Raphoe, Kells, and elsewhere, and they belong essentially to the period when the Northmen made their attacks on Ireland. Nevertheless, they continued to be built until the middle of the thirteenth century[298]. All these facts are of vast interest when we recollect those Lincolnshire churches which, folk-memory tells us, once served as refuges from the Danes.

Normally, the round tower contained three stories, and, like many of the English towers before described, had the outside door placed high up in the wall, so that it was accessible only by means of a ladder. Similarly, the oratory of Kells has three small chambers, or attics, between its round barrel vault and the outer pointed roof of stone; these chambers could be reached only by the aid of a long ladder, and entered through a hole in the inside roof. Viewed in comparison with the other buildings of the period, the oratory represents a lofty structure. It has a highly pitched roof, and receives no light save what is afforded by two small windows. This building is assigned by Miss M. Stokes to the year A.D. 807[299].

In passing, we note that the round tower was probably developed from the beehive hut, and became specialized for a definite purpose, which we shall discuss in a moment or two. The beehive house, in its turn, was most likely evolved from the wattled dwellings of the ruder aborigines of Ireland. Professor A. C. Haddon cited authorities to show that the round towers structurally betray their pagan design, by their retention of string-courses which serve no useful object. The towers, he considers, are, in reality, derived from primitive wicker huts, circular in plan, and of a somewhat tall type[300].

We now inquire what purpose the towers were intended to serve. The answer, though given with a fair degree of decisiveness to-day, has been long in coming, and there still exists a minority of writers who dissent from the orthodox view, and who favour other theories, some of which are of the fantastic kind.

The earliest name given to the round tower, so far as is yet recorded, is cloictech, or belfry (A.D. 948). Naturally, therefore, one is led to associate the towers with bells; yet, seeing that only small hand-bells are believed to have been in existence at that date, a large tower of masonry would not be needed for their reception. Nor, again, would these small bells have been well heard, when rung from the top of the tower, even by folk standing at the base[301]. The insignificant windows of the earlier towers are, for some unknown reason, usually near the floor. The later structures, it is true, have windows near the top, facing different points of the compass. These examples may have been used as belfries proper. Speaking generally, the towers were, nevertheless, not belfries. Even could it be proved that, co-eval with the towers, there existed a knowledge of bell-casting, that would not settle the question, because there are no structural signs that bells were ever hung, or rung, within the buildings. We may conclude, then, that the towers were, in part, depositories for bells, though not themselves actually bell-towers.

Dismissing a number of unsupported speculations, one by one, we are finally shut in to the conclusion that the towers were originally defensive. They were refuges for men, and storehouses for valuable property. This is the verdict delivered by those who, like Miss M. Stokes and Miss E. Hull, know the round towers best[302]. Mr G. T. Clark, in his early study of the subject, thought that the towers were principally intended to receive bells, and that their uses for refuge and storage were later adaptations. In after years, he changed his opinion, and declared that what he had formerly deemed the secondary purpose, was in reality the sole one[303]. A more recent authority, Mr Francis Bond, has compared the round tower with the Italian campanile, of which he considered it to be a local variety. But he draws this great distinction: the Italian structure was built to receive bells, and may afterwards have served for defence; with the builders of the Irish towers, refuge was the primary idea, though, at a subsequent period, bells were hung in the fortress. There remains an authority whose voice must be heard with great respect. Mr J. Romilly Allen, while agreeing that the bells of the early Celtic church were portable, and that they were rung by hand, contends that the towers were erected to accommodate bells of a heavier kind. Nevertheless, he admits that the Viking invasion gave an impetus to the building of the towers[304].