Many years ago, the commanding position which the village church frequently occupies forced itself upon the attention of the writer. As will be shown hereafter, the builders, for some cogent reason, which may yet be determined, chose a spot having considerable natural advantages with respect to strength and security, and there they erected their temple. These geographical observations would not alone have been sufficient to evoke a general theory, had not other facts gradually come into view. One of these facts was the frequent association of the church with earthworks, tumuli, and similar relics of antiquity, and it was this conjunction which raised the inquiry whether the relative positions could, in all cases, be merely accidental. A closer and more prolonged study, involving much personal investigation, together with a review of many isolated fragments of archaeological literature, led to the conclusion, almost irrefutable, as it now appears, that many of our churches stand on pagan sites. A secondary deduction from the observed facts was the probability that, in some cases, there has been almost continuous site-occupancy since the first Christian church was reared.

During the inspection, numbers of records, based on imperfect knowledge or on speculations of the earlier antiquaries, have had to be discarded; in other instances the test has been successfully borne. The presentation of the evidence, with its length of detail, may be somewhat wearisome to the reader, who may, however, console himself with the thought that he has escaped at least a moiety of the mass which has been winnowed. Furthermore, one may recall the truth set forth by Professor E. B. Tylor when apologizing for wealth of detail in stating a case: “The English mind, not readily swayed by rhetoric, moves freely under the pressure of facts[1].” One may, for a moment, arouse interest by a new hypothesis, but it is only by the accumulation of facts that public opinion is perceptibly influenced in the end.

Viewed strictly, every Christian church was originally built on a pagan site, but we will limit the meaning of the adjective so that it shall apply to those churches which were erected, not on virgin soil, but on some spot once devoted to heathen worship, whether beneath a roof or under the open sky. This definition would narrow the scope of the inquiry; nevertheless, to arrive at a clear decision we shall have to survey the whole question from pre-Roman times onward.

Our path will be greatly cleared if we recognize, and remember—what is too commonly forgotten—that there was a Christian church in Britain long before the mission of Augustine in A.D. 597. Apart from legends, and documents of doubtful authenticity, some writers claim to have proved that British Christianity was well developed before the close of the second century of our era[2]. Other authorities assert that the evidence for the second century is unhistorical, and that the first genuine reference to Christians in Britain is made by Tertullian (c. A.D. 208)[3]. However this may be—and the question of the exact date of the introduction is foreign to our present study—there is unanimity as to the existence of a strong British Church soon after the death of Constantine (A.D. 337). It is even stated that, at the date just mentioned, Britain was as fully Christian as any country in Europe[4]. At any rate, it is beyond dispute that, in A.D. 314, the British Church was represented at the Council of Arles, in France, by three bishops, together with a priest and a deacon[5]. Certain writers go further, and contend that, before Britain was cut off from the Empire, the Church had a vigorous corporate life of its own[6]. How long this organization endured, and to what extent it was weakened or shattered by the shock of the Teutonic invasion, are more debateable subjects. It is possible, however, that a remnant of churchmen survived to greet the advent of Augustine[7]. This only must be said, that the existence of any continuity of Christian tradition, however slight, might render the task of deciding what is a pagan site more difficult. Under the influence of an unbroken tradition, churches might be constantly rebuilt on the old foundations; hence, if this assumption be made, additional testimony would be necessary in order to establish the theory that any original structure was set up by the heathen. If such evidence were lacking, the successive buildings would simply strengthen the hypothesis of continuity of Christian worship, but would leave untouched the problem of heathen sites.

The first problem to be attacked, then, concerns the existence of Christian churches during the Roman period, and the after-history of such buildings. Do any of these churches remain to us? The available evidence seems to show that, in outlying districts, at least, churches were constructed of wattle, and, of these structures, not a wrack could possibly have persisted until the present day. In the cities, more durable materials, limestone, flint, chalk, and baked tiles, would be employed, and there is some likelihood that portions of buildings so constructed would successfully resist the ravages of vandals and the fury of storms. Now, it is singular that the churches which will least stand the critical test of the architect and the antiquary with respect to a Roman origin, are precisely those which the popular vote declares to belong to that period. The churches thus misunderstood are those which have large quantities of undoubted Roman materials built into their walls. The catalogue is of formidable length, but may be soon dismissed after a few typical examples have been noticed. The walls of the cathedral church of St Albans contain abundance of Roman material, and a continuity of buildings, dating from the Roman occupation, has therefore been hastily assumed. Bede, it is true, relates that a church was built over the grave of St Alban at Verulam[8], and it is possible that the spot is now covered by the cathedral, but we cannot wisely go beyond this, especially when we remember how plentiful were the Roman materials close at hand. The fact remains: from the time of the erection of the memorial church to the founding of the monastery in A.D. 793, we have an interval which is unbridged by trustworthy testimony. A generation ago, Mr Roach Smith, a most sagacious observer, compiled a list of Kentish churches which he thought might be probable restorations of pre-Saxon structures[9]. In all of these Roman materials were found. Some of the churches, however, like those of Reculver and Lyminge, had peculiarities of site, and these examples will be noted later. Among the Kentish churches whose “Romanity,” as the early antiquaries would phrase it, must be discredited, are those of Burham, Leeds, Southfleet, and Lower Halstow. Yet the last-named church is chiefly built of Roman spoil. The “Garden County” also yields Cuxton[10] and St Paul’s Cray, with many another church inwrought with Roman tiles. Crossing the Thames estuary, we find, according to Mr Guy Maynard’s computation, thirty-five Essex churches which have Roman tiles in their walls[11]. A writer in the Athenaeum, commenting on this list, gives a higher figure, and asserts that Essex contains at least sixty such churches[12]. We may safely infer from these facts that Roman ruins existed in the neighbourhood of each of the sites at the time when the walls were built. Any further conclusion must be viewed with suspicion, unless Roman remains are discovered beneath the buildings. The “argument from silence” is beset with peril in any department of archaeology. Moreover, some of the churches in the list—which might be greatly extended—belong, as Professor Baldwin Brown has observed, to purely Mediaeval settlements, and consequently have little evidential value[13].

We turn to a different class of churches—those which occupy the sites of Roman villas. The importance of these examples rests on the probability that some of the wealthier Roman converts would allow their dwelling-houses to be consecrated for Christian worship. From a small reception-room, arranged like an ordinary church, there might be developed a Christian building, with chancel, nave, and aisles complete. A scrap of testimony, slight though it be, favours this hypothesis. It is the discovery, on a mosaic, among the ruins of a Roman villa at Frampton, Dorsetshire, and again on a tile from the villa at Chedworth, Gloucestershire, of examples of the Chi-Rho monogram[14]. This sacred monogram has also been met with on such objects as bowls, seals, and rings. Seeing that the symbol was not used in Rome before A.D. 312, its presence in Britain cannot date earlier. On the other hand, remembering that the Roman departure took place in A.D. 410, we can scarcely assign the Chi-Rho to a later date. Mr J. Romilly Allen is therefore plainly near the truth when he attributes the British examples to the late fourth century[15].

The validity of the evidence afforded by the Chi-Rho, while unquestionable so far as the existence of British Christianity is concerned, is not decisive with respect to site-continuity. At the outset, one demands that the monogram should be found in juxtaposition with the later Christian churches built on older sites—not isolated from such buildings. On the other hand, it would be passing strange if a large number of churches came to be built by chance on, or adjacent to, the areas once occupied by Roman villas, whether the confirmatory Chi-Rho were discovered or not. If we consider the case of direct continuity non-proven, and yet rule out the possibility of accident, a choice of two theories seems to be presented. We might either suppose that the church builders were keenly anxious to utilize ruined villas, or that, believing those villas to have been centres of pagan family-worship, deliberately chose to set foundation over foundation. That this second alternative is not altogether fanciful will be seen hereafter. A few examples of villa sites will now be given.

The churches of West Mersea, in Essex, and Wroxeter, in Salop, are believed to stand on sites of Roman villas; a little contributory testimony is afforded by the fact that the shaft of the font, in each case, is fashioned from the drum of a Roman column[16]. In the case of Wroxeter, however, the only tessellated pavement recorded by Professor Haverfield was found a little to the north of the church. The conditions are supposed to have been similar at Haydon and Chollerton, in Northumberland, and at Great Salkeld, in Cumberland; in all of these instances the fonts are said to be hollowed out of Roman altars[17]. During the rebuilding of St Swithin’s Church, Lincoln (A.D. 1880-88), a Roman altar (Fig. 1) was discovered beneath the tower. The old fabric belonged to the Decorated period, while the altar dates from the second century of the Roman occupation. There is thus an intervening space of more than a thousand years, and this gap cannot yet be actually bridged over. At the deserted church of Widford, in Oxfordshire, portions of a Roman tessellated pavement were found in the chancel[18].

Professor Seebohm, who closely studied the district around Hitchin, and discovered strong proofs of unbroken occupation of village sites, gives some interesting examples which bear on our subject. He thinks that the church of Much Wymondley, near that town, stands within a Roman holding, probably that of a retired veteran[19]. A Roman cemetery was discovered hard by, and to the east of the church is a double “tumulus,” which Professor Seebohm conjectured to be a “toot-hill,” or a terminal mound[20]. These toot-hills will be again mentioned; meanwhile, we are bound to notice that more recent investigators claim this particular hillock as an early castle-mound. Nevertheless, it is stated that the mound and its associated bailey-court have been

inserted into the corner of a larger (and presumably earlier) rectangular work[21]. A Roman villa is recorded from a field near Litlington churchyard, Cambridgeshire, and a Roman cemetery from a spot a short distance away[22]. Other examples have been noted at, or near, the churches of Woodchester and Tidenham, in Gloucestershire, and Wingham, in Kent[23]. The first-named instance is the most instructive. In the churchyard an inscribed pavement, 25 feet in diameter, was uncovered, and near at hand, the ground plot of an extensive building was traced. The neighbourhood of St Saviour’s Cathedral, Southwark, has yielded quantities of Roman remains. A portion of a pavement is shown in Fig. 2. Within the last two or three years, Roman pottery, and the upper portion of an amphora, have been discovered while alterations were being made. These relics may be seen in the south transept. Whether the long list of “finds,” given in the Victoria History of London, justify the old tradition of a pagan temple may be doubted, but, at least, the former existence of a villa is indicated. A tessellated pavement was discovered in the south transept of Southwell Cathedral, and Mr Francis Bond conjectures that this relic may have belonged to a Romano-British basilica which existed there in the third century. Did such a building exist, the church which St Paulinus is believed to have founded on this spot in the seventh century had a prototype, which dated four hundred years earlier[24]. In his recent standard work on Westminster Abbey, Mr Bond has also recorded the finding of a portion of a Roman wall, in position, under the nave of the Abbey, and a Roman sarcophagus in the northern part of the nave. Roach Smith alludes to foundations, probably Roman, which were unbared at Chalk Church in Kent[25]. The Saxon church of Bosham, Sussex, is another claimant for superposition on a Roman villa[26], and the fine old Saxon building at Brixworth, Northants (Fig. 3), is a further example, although no part of the present structure is older than the eighth century[27]. Our list is by no means exhausted. A very fine mosaic floor, worked in seven colours, together with a bath and other remains, were laid bare many years ago at Whatley House, Somerset, just behind the ancient church of Whatley. When the church of St Mary Major, Exeter, was being rebuilt in 1866, the Norman foundation was seen to cover a Roman tessellated pavement[28]. Still more recently, in 1906-11, during the process of underpinning Winchester Cathedral, the workmen discovered Roman coins and tiles[29]. These remains may have had no causal

connection with the present building, or with any hypothetical predecessor, yet the discovery was curious. We need have no desire to strain the evidence. In such instances as Winchester and Wroxeter, Roman ruins and Roman sites would be so plentiful, that no enterprising Saxon builder would overlook the economical value of the spoils. Again, he might unwittingly select an old site concealed by long-continued labours of earthworms, and by natural agencies of weathering. Yet even this admission will, in its turn, react if accepted too eagerly or too fully. We are dealing, so far, primarily with the existence of early British churches, and if we urge that old sites were re-occupied unintentionally, because they lay hidden from view, we imply that, in other cases, foundations hitherto undiscovered may rest beneath later architectural monuments. In other words, the foundations of a pagan temple may lie beneath a Mediaeval church. There may have been continuity up to a certain date, and then a break; after which a new builder started work over the forgotten floor. Seeing that most of the Romano-British towns, at least, were continuously occupied since their first establishment[30], and that, as already shown, old material was intercalated between the courses of masonry in newer buildings, these facts alone would be sufficient to account for the obliteration of the earlier work[31].

Having now referred to the very doubtful instances of continuity represented by fabrics in which there has been an adaptation of Roman materials, and having glanced at those churches which stand on the sites of earlier buildings, we turn to Christian edifices which have been built adjacent to Roman camps. At present, we will consider those cases in which there is actual contiguity, but only a suggestion of purposiveness. The ivy-clad church of Ashtead, in Surrey, stands within a rectangular earthwork, partially defaced, and the visitor will readily detect Roman tiles in the walls of the chancel. At Rivenhall, in Essex, tesserae and Roman pottery were dug up in the churchyard, and a villa was unearthed in the neighbouring field. From the data available, one cannot decide whether or not a camp is indicated[32]. In the same county, we notice Stoke-by-Nayland, while Suffolk supplies us with the camp Burghcastle—a most interesting example. St Furseus, or Fursey, built a monastery at this spot, but there remains only the church, which lies a little to the north of the Roman fortifications. Its walls contain triple bands of flints, faced by Roman workmen, while vases and potsherds have been discovered in the vicinity[33]. Squared flints of Roman workmanship were also found at Caister by Norwich[34]. The church of St Edmund, at the last-named village, was built by Mediaeval architects at one corner of a Roman earthwork, which encloses an area of 34 acres. The present church, as Professor Haverfield points out, is certainly not a Romano-British “sacellum” or temple[35], but, in the absence of excavations, one cannot assert that no earlier ruins lie buried underneath the edifice. The oft-quoted instance of Castle Acre, also in Norfolk, must be dismissed as spurious. Professor Haverfield, who has carefully examined the evidence, could find no proofs in support of the tradition of a camp, though there was evidence of Roman occupation in the neighbourhood[36]. Under the present section, however, we must include Market Overton and Great Casterton in Rutland. The church of the latter village is situated at the south-west angle of an earthwork, presumably Roman, though of earlier construction than the Roman road hard by[37]. At Market Overton, the church stands entirely within a square Roman camp[38]. In the adjoining county of Lincolnshire, we get examples at Caistor and Ancaster[39], places bearing tell-tale names. The church of Horncastle is within a few yards of a Roman wall, a portion of which remains visible above the land-surface[40]. Lincoln Cathedral is built partly within and partly without a Roman camp[41].

In Durham, the church of Chester-le-Street, which contains some traces of pre-Conquest work, was originally inside a Roman camp, now unfortunately destroyed[42]. Ebchester Church, also in Durham, stands at the south-western corner of the ancient Vindomora, and has a foundation of large squared stones, but little can now be seen of the surrounding fortifications[43]. While surveying the North of England, we notice Moresby, near Whitehaven[44]. In Scotland, to mention but one case, we have the Cistercian Abbey of Cupar-Angus, which was built, in A.D. 1164, within the boundaries of a Roman camp[45]. Returning to the south, we discover, in the churchyard of St John’s-sub-Castro, at Lewes, a small Roman camp, of which the vallum is still traceable[46]. Porchester, in Hampshire, is a square-walled fort which occupies an area of 9 acres, and which encloses a Mediaeval keep and bailey-court at the north-west corner, and a Mediaeval church and graveyard at the south-west corner[47]. In like manner, the Norman church at Silchester nestles within the celebrated Roman settlement. Here our list of Christian churches placed within Roman camps must be curtailed, for we have still to consider earthworks belonging to an earlier period. The reason for separating the two classes of earthworks is, that those churches which were reared within Roman camps may, probably, in some cases, have replaced more primitive buildings, while those built inside prehistoric forts most likely had no predecessors. In other words, we shall have to search for different motives inducing the choice of the two respective series of sites.

At the very threshold of the inquiry a marked difference is noticed: the pre-Roman earthworks contained no building material to entice the churchmen within their boundaries. Turning to individual examples, we find a most instructive case at Knowlton or Knollton, Dorsetshire, four miles south-west of Cranborne. Here, a ruined church built by Norman labour, though not necessarily representing the first church reared on the spot, stands within a round British earthwork (Fig. 4). The ditch, or fosse, of the enclosure is situated on the inner side, as in the renowned earthwork at Avebury, Wiltshire. The Saxon church at Avebury dates in the main, perhaps, from the early tenth century, and stands just outside the vallum. Some writers have inferred, from the presence of the inner fosse, that these enclosures had religious, or, at least, sepulchral associations. The Knowlton earthwork is one of a group, and close by is a cluster of ancient, storm-beaten yews[48]. Such a collocation, as will be seen in Chapter IX., is not without significance.

Another dilapidated chapel, now used as a barn, is situated within the oval camp of Chisbury, near Great Bedwyn, Wiltshire. This earthwork, which has double, and in some parts treble, lines of trenches, is described by Sir R. Colt Hoare as one of the finest specimens of castrametation in England. One rampart is 45 feet in height. The existing ruins represent a Decorated fabric which was dedicated to St Martin, but Mr A. H. Allcroft, in his Earthwork of England, suggests that a church was erected here after the drawn battle between Wessex and Mercia in A.D. 675. On the hill above Standish Church, Gloucestershire, is a somewhat notable camp. Although it is said that the ditches were deepened during the Civil War, and although Roman coins have been dug up in large numbers[49], it is conceived that the camp was originally British. On the height just above Gunwalloe Church, Cornwall, is a “cliff castle”—one belonging to the Group A, as defined by the Congress of Archaeological Societies in 1903[50]. Such earthworks are inaccessible along a portion of their boundaries, on account of the presence of cliffs or water. The site of the church of St Dennis, also in Cornwall, is associated with a “hill castle[51],” which is assigned to the Group B. In this class, the earthwork follows the contour of the hill. Another contoured camp, much disturbed and defaced, is situated on St Anne’s Hill, near Midhurst Church, Sussex[52], while a small circular fortification may be seen to the west of the churchyard of South Moreton, Berkshire[53]. Coldred Church, Kent, was built actually within a fortress, conjecturally of Romano-British date[54], though the elevation of the earthwork is rather exceptional for that period, being about 370 feet above the sea-level, and 50 feet above the valley towards the west. Again, at Kenardington, also in Kent, an earthwork of unknown age, now much mutilated[55], surrounded the graveyard and part of the neighbouring fields.

The so-called Dane’s Camp (Group B) at Cholesbury, Bucks., 600 feet above the sea-level, encircles the church of St Lawrence with its embankment[56]. Another St Lawrence, at West Wycombe, in the same county, is built inside a ring earthwork (Group B), which crowns the hill. This fort, probably of British construction, is remarkable for its double-terraced defences, and for the manner in which it commands three converging valleys[57]. A somewhat similar example was once visible at Brownsover, near Rugby, where, a century ago, the church and village were enclosed within elaborate entrenchments. These represented a fortress, constructed on a ridge which overlooked the valleys of the Avon and the Swift. The fort was probably prehistoric, although a cinerary urn, found in the churchyard, was identified as Roman.

The hill-village of Burpham, in Sussex, is clustered near an oblong promontory fort (Class A) constructed on a tongue of land, around which a loop is formed by the river Arun. A gigantic vallum and exterior fosse cross the neck of the peninsula. The early Norman church of the village stands but a few yards beyond an entrance breach in the northern rampart. Mr A. H. Allcroft, pursuing the “method of exhaustions,” declares the earthwork to be Danish, and Mr P. M. Johnston suggests that the church occupies a pagan site. At all events the juxtaposition can hardly be considered casual.

Immediately to the east of Hathersage churchyard, Derbyshire, may be seen a simple circular earthwork, consisting of a high rampart with a moat outside. It is classed by Dr J. C. Cox in the division C of the scheme above-mentioned[58], namely, the division which embraces round enclosures of a defensive character. An analogous earthwork adjoins the churchyard of Tissington, also in Derbyshire[59].

Without pursuing this quest further, one or two pitfalls must be pointed out. Entrenchments found near a parish church may sometimes represent portions of the “ring fence” of a Mediaeval settlement; and the banks, which once bore a hedge or palisade, might be hastily ascribed to an earlier period. Mr Allcroft, in the work just mentioned, cites numerous warning examples. Again, banks of boulder clay or glacial drift may assume a false appearance of ridging, as if due to the work of man. To glacial action I venture to assign the surface irregularities near Ludborough Church, Lincolnshire, though they may represent the partially erased banks of the Mediaeval village. Close by the neighbouring churchyard of St Lawrence at Fulstow, one sees similar unevenness of the ground, the most important hillock being perhaps a grave wherein were buried some sixscore parishioners who died of the sweating sickness in the early seventeenth century. Once more, the traces of earthwork, military or agricultural, below the church of St Michael, on Glastonbury Tor, Somerset, may not be very ancient, and I should not connect them in any manner with any ideas which were held by the Gothic architects.

We next inquire why churches should have been built in situations such as those which we have been considering. Mr Allcroft, arguing apparently from the assumption that the church was a defensive building—in fact, almost the only one in the parish—considers that it was sometimes built near earthworks for additional security[60]. That Mr Allcroft’s premises are sound, I shall attempt to show in the next chapter. That, in exceptional cases, his conclusion is correct, one would not care to deny. But can the theory be of general application? Scattered throughout the land are churches built in exposed and lofty situations, so that traditions, varying in detail, but related in their main principle, have sprung up to account for the choice of these isolated and inconvenient positions. Most of the stories put fairies, or, more commonly, the Spirit of Evil, in opposition to the efforts of the builders. Churches were moved in a night, or the day’s work was undone by the malignant foes. In cases of this kind, as in those instances where churches stand in some secluded meadow, the reason may occasionally be found in the churlishness of the manorial lord, or in the fact that the village settlement has shifted since the church was built. Houses are demolished and rebuilt, but the church remains. The desire to place the church in an impregnable spot may more frequently account for the hill-structures, which will be considered in Chapter III., though not for the churches near earthworks, nor for the sequestered churches in the fields. Some other explanation must be sought, and, curiously enough, Mr Allcroft has incidentally suggested two other theories. The early missionaries to the pagan Saxons, he supposes, made their headquarters on deserted Roman sites, first, to demonstrate their own power in successfully defying the evil spirits which haunted those spots, and secondly, through the bad reputation of these earthworks, to obtain “something of a guarantee against molestation by human beings quite as formidable[61].” While not agreeing that the second motive would be very influential, with the first suggestion I find myself more in harmony. The miraculous power of withstanding devils and demons would not be without its effect on the ignorant. Moreover, the claim would be as effective during the Mediaeval as it was during the Saxon period. For we are not to suppose that superstition fled the land on the advent of the Normans. Who were these new folk, and what were their antecedents, that they should be free from slavish fears of the unknown? Legends were without doubt attached to prehistoric remains down to a late date; how intense and how gross are the superstitions of country folk even in our own day, only the close student of men and books can be aware. Thus, for some reason, inexplicable, except on anthropological grounds, there exists among the Lincolnshire woldsmen a prejudice in favour of burial on the heights, and many similar facts could be given.

Above all these causes of selection of prehistoric sites, however, one may place the spirit of compromise which actuated the missionaries. Everywhere, the preachers found that the Saxons, who were unaware of the real origin of the old defences, attributed them to diabolism. Devil’s Dykes, Devil’s Highways, Devil’s Doors, as has been shown in another volume, meet us in every part of the country[62]. Believing firmly in the diabolic origin of the earthworks and megaliths, the Saxon was moved to fear, and to that slavish respect which is the child of fear. Yet it was pre-eminently in the open country, where such objects abounded, that the Saxon loved to dwell. It has been shown that, however much he may have avoided the walled towns—and these he did not shun altogether—the Saxon settler had no antipathy to occupation of the deserted villas and rural settlements[63]. Here, then, the potential convert, with his superstitions and aversions, lived and toiled. The monuments of earlier races he regarded with sacred awe. It would be well-nigh impossible to wean him from his creed by direct denunciation; it would be easy to win him over by toleration and compromise, and this possibility seems to supply the real explanation why earthworks and other spots with weird associations were chosen for many of the early churches. If it be asked why still more instances are not forthcoming, it may be answered that the earthworks were frequently too remote from settlements on the plains, and were too elevated in position, to tempt the builders, even when the desire for protection reinforced the primary purpose. Moreover, though the earliest open-air preachers in Saxon times may have selected the earthwork as a pulpit, the permanent church would not necessarily be built within that area. (It will save misapprehension, if an explanation of the use of the word “Saxon” be interpolated here. In strictness, there is a clear distinction between Angle and Saxon, dialectically and archaeologically. But it is impossible always to observe the differences, especially when the data are scanty. The term will be employed, then, in its old loose signification, to denote, as Mr Reginald A. Smith says, “the roving Teutonic bands that for centuries infested the Northern seas.”)

We have now glanced at those churches which contain remnants of Roman ruins, and others which are built over Roman villas, or within Roman camps, and we have been led insensibly to examine buildings which are connected with earthworks of other ages. The problem of site-continuity has constantly impinged upon the question of continuity of fabrics. A few paragraphs may now be devoted to a consideration of those churches which lay claim to a possession of one or both of these features. The small ruined church of St Mary, within the confines of Dover Castle, is a well-known example. It stands in juxtaposition with an octagonal structure, usually described as a pharos, or lighthouse (Fig. 5), and believed by some to be a fort belonging to the Romano-British period. This polygonal tower has an exterior casing of flint, dating from the fifteenth century, but the original uneven masonry of rubble and flint, bonded with bricks at intervals, is still visible at the base. The supposition is that the church, with the lighthouse, was utilized for Christian worship during Roman times. By most modern authorities, the church itself is attributed, and perhaps more correctly, to the late Saxon period[64]. Lyminge, in Kent (p. 4 supra), is another claimant. The foundations of a seventh-century chapel, probably of apsidal basilican plan (Fig. 6), have been traced here (A.D. 1899), but it is supposed that the present church, though rich in Roman materials, belongs entirely to a later epoch[65]. At Reculver (Regulbium), near Herne Bay, there is an example of a church which Professor Baldwin Brown places with that of Dover in a distinct category as representing possible authentic relics, since the buildings stand alone within deserted Roman stations. The church at Reculver stands over the foundations of a basilica, but the present building is probably altogether post-Roman, the earliest known date for the existence of a church on this spot being A.D. 670[66].

Dean Stanley held the belief, once shared by many antiquaries, that in St Martin’s at Canterbury we have a veritable monument of early British Christianity—a monument, moreover, erected over a pagan temple[67]. Bede asserts that there

was a church on this spot in Roman times, and that the building which existed in his day retained relics of the older structure[68]. In spite of this tradition, the popular belief is only doubtfully tenable. The site is old, and there may have been unbroken continuity, but the present building, though doubtless largely composed of the original materials, has been altogether re-arranged[69] (Fig. 7). An exception may perhaps be made for portions of the western nave, which Professor Baldwin Brown considers may represent early work. St Pancras, at Canterbury, by some writers judged to be older than St Martin’s, must, under reserve, be given up, for similar reasons. Foundations, nearly complete, of a single-celled apsidal church have been revealed to the excavator, but the actual persistence of work above the surface is not demonstrated[70]. Other churches put forward are Ribchester, in Lancashire[71], and the chapel of St Peter’s-on-the-Wall, at Bradwell (Othona), in Essex. St Peter’s Chapel represents a barnlike building, of which the materials were evidently quarried from the adjacent fortress, but, once again, proof of continuity is lacking.

St Joseph’s Chapel, at Glastonbury Abbey, presents us with an interesting case of probable retention of site, though not necessarily of continuous buildings. The earlier history of Glastonbury is, unfortunately, mainly a history of legends and traditions. We may well discredit the tale, told by the imaginative William of Malmesbury, a millennium after the alleged event, that, so early as the first century of the Christian era, a chapel constructed of osiers existed at this spot. That some kind of primitive church or oratory, with walls of wattle, and a roof of reeds, was set up during the Roman occupation is, however, very probable, and it may fairly be supposed, though it cannot be proved, that no break had occurred when the Saxon abbey was founded[72].

Among other churches for which a reasonable claim has been advanced is that of Jarrow, which Professor Brown places in his period “A,” that is, the period anterior to the year A.D. 800[73]. Again, the oldest part of the cathedral church of Canterbury, as attested by experts, slightly supported by Bede’s description, may be a relic of Roman Christianity[74]. We pass from these examples in order to glance at a church whose age may now be deemed undisputed, namely, the small apsidal church or basilica which was uncovered in 1892 at Silchester (Calleva Atrebatum). The nature of the building was at first much canvassed, and some authorities, relying chiefly upon the absence of Christian symbols in the mosaics, and upon other details, denied that the foundations were those of a church[75]. Curious to relate, the Chi-Rho, along with the Omega, was found impressed on the side of a small leaden seal which was dug up in the Silchester Forum[76], hence, if the basilica has yielded no evidence of Christian symbolism, such testimony lay hidden at no great distance. To be brief, not only is the basilica now accepted as genuine by the best authorities, but Messrs G. E. Fox and W. St John Hope declare that it is the only example of a Christian church of Roman date yet found[77].

On the ruins of the Silchester basilica no Gothic church sprang up, so that there was not site-continuity. Yet the parish church, which was afterwards built during the twelfth century, within the enceinte of the destroyed Roman town, has a direct bearing on the subject of this chapter. It is one of the two instances of churches which Professor Brown admits as having possibly superseded pagan Roman buildings. He does not, however, concede that we have any examples of Saxon churches which once actually formed parts of such temples. In all cases, the form and orientation of the churches, he asserts, betray an ecclesiastical origin. The churches may point to a survival of Romano-British Christianity, but that is another question[78]. Nevertheless, as Professor Brown himself notes, Silchester parish church was built close to the remains of two small Roman shrines of Gaulish type. The orientation of the church exactly agrees with that of one of the shrines, and this may indicate some relationship[79]. Messrs Fox and St John Hope have stated that the list of edifices dedicated to pagan deities in this country is very scanty, yet it is noteworthy that, of this list, three were recorded at Silchester. Moreover, one of these temples was found lying partly under the graveyard of the parish church, and partly under the buildings of an adjacent farm[80]. “Perhaps,” say these writers, “the rising power of Christianity, as seen in the little church [the basilica] within the south-eastern corner of the Forum, may have made for their destruction[81]” [i.e. the destruction of the shrines]. May we not add that, should someone excavate a second Silchester, further evidence of this kind might be obtained? Dr Thomas Ashby’s explorations at Caerwent (Venta Silurum) have, so far (1910), yielded no certain traces of a Christian church. The basilica discovered on the north side of the Forum is of a civil, not religious, character. We might frankly discard all the examples previously given, save perhaps those in which churches stand over Roman villas, and yet come to a wrong conclusion by arguing from the absence of particular witnesses. Other deponents may press forward. Before, however, we can examine these, we must make a rather lengthy digression to inquire if there exist a priori reasons for the annexation of pagan sites by Christian teachers.

We have, in proceeding to this examination, principally to consider the policy which was pursued by the early missionaries. Writing about Christianity in general, Harnack has shown that, during the third century, it united enthusiasm with the spirit of tolerance. “Stooping to meet the needs of the masses,” the leaders studied polytheistic customs, instituted festivals and saints, and utilized sites already deemed sacred. To express the fact otherwise: the religion became syncretistic in the proper meaning of that term[82]. Christian and pagan ideas were blended. Following the wise, and, indeed, the only practicable method—that of peaceful permeation—the Church often retained the forms of heathen ceremonies, while actually investing these with new meanings. The process has been pithily expressed by Sir G. L. Gomme: “Christianity was both antagonistic to, and tolerant of, pagan custom and belief. In principle and purpose it was antagonistic. In practice, it was tolerant where it could tolerate freely[83].”

As a matter of history, however, we learn that the policy did not remain strictly consistent, and a struggle for survival ensued. Under the rule of Constantine, the tendency was to destroy heathen temples and their idols, but by the Edict of Theodosius (A.D. 392), pagan shrines were to be dedicated as Christian churches. Later, the Edict of Honorius (A.D. 408) definitely forbade the demolition of heathen temples, at least in the cities[84]. These enactments seem to have a direct bearing on cases like that of Silchester and upon other examples, to be described hereafter. Leaping over a gulf of nearly two centuries, we discover Pope Gregory the Great (A.D. 601) sending a letter to the Abbot Mellitus, who was then about to visit Britain, commanding that, while idols were to be destroyed, the temples themselves were to be preserved. Holy water was to be sprinkled in the buildings, altars were to be erected, and sacred relics were to be placed therein. Anniversary festivals were to be appointed, and the new worship inaugurated[85]. Keeping this in mind, we are not surprised to find that, on the conversion of Ethelbert, two or three years previous to the Gregorian edict, Augustine received a licence to restore, as well as to build churches[86]. Whether these churches were pagan temples which had been partially despoiled, or Romano-British basilicas which had fallen into decay, we are left to conjecture.

On the continent, the breach of continuity of policy was still less perceptible. Grimm distinctly states that churches were erected on the sites of heathen trees or temples. He warns us against false conceptions of history. We are not to picture the poor peasants as being ruthlessly expelled from their accustomed places of worship. The heathen, he declares, were not so tame and simple, nor were the Christians so reckless, as to lay the axe to sacred trees, or to fire the pagan temples. The rude forefathers of the hamlet trod the old paths to the old site. Sometimes the very walls were retained, nay, the local idol or image was retained outside the door or within the porch. Thus, at Bamberg Cathedral, in Bavaria, zoomorphic stones, inscribed with runes, passed the examination of lenient judges[87]. Again, pagan festivals were converted into Christian holy-days. The Yule-tide merry-makings in honour of Thor—revels which have also been connected by some writers equally with the gods Adonis, Dionysos, and Mithra—became the festival celebration of the birth of Christ. Canon E. L. Hicks (now Bishop of Lincoln) contends that the observance of the exact date, December 25th, as Christmas Day, is directly borrowed from Mithraism[88]. The old German feast in memory of departed warriors was metamorphosed into All Souls’ Day, when the spirits of resting believers were kept in mind[89]. As with holy-days, so with symbols. Thor’s hammer was replaced by the Christian Cross, and the heathen sprinkling of newly-born babes became Christian baptism[90]. Thus, by the retention of holy oaks, of idolatrous feasts, of pagan symbols and ceremonies, of the heathen names for the days of the week, the new religion gained entrance. In Ireland, where the problem to be faced was remarkably complex, the Christianization of pagan myths was very noticeable[91]. Here, the very names of the feasts long continued as in pagan times. Only when the conciliatory policy had “eased the yoke of the new ordinances,” was it possible to take drastic measures, and to extrude heathenism from the places of worship[92].

But this time was slow in coming. In the heart of the Empire, as Friedlander has shown us, the triumphant Christians did, indeed, assimilate many heathen practices, yet they strove hard to stifle paganism altogether. On the other hand, all over Northern Europe, the spirit of compromise was at work. In Sweden, during this transition period, old associations were so strong that prayers to Thor and Freya were often mingled with Christian orisons[93]. Professor F. Kauffmann speaks of the great temple of Upsala, with its evergreen tree, and its mysterious sacrificial well, which received the bodies of the slain. So late as the eleventh century, this temple still stood in all its splendour[94]. Professor O. Montelius, while noting the frequency with which sacred stone-circles are associated with the church, considers that the cromlechs were not places of sacrifice, but of judgement. This idea is gaining ground in England, where also there is a tendency to change the nomenclature of megaliths. (To avoid confusion, it must be noted that “dolmen,” in these chapters, refers to a “table-stone,” that is, several upright stones capped by a flat one. “Cromlech” is used in its Breton sense of stone-circle, not in Welsh and Cornish significations of table-stone, nor in Sir Norman Lockyer’s restricted connotation—a kind of “irregular vault generally open at one end.”) At Gamla Upsala, near Upsala, a church was built on the site of a temple, which was the traditional burial place of Odin, and the centre of his worship. Modern excavations at this spot have yielded bones of horses, pigs, and hawks, together with relics of gold and silver[95]. This example is instructive, alike for its testimony to the value of folk-memory, and for its illustration of the employment of a pagan site. But, indeed, example can be piled upon example. At the Danish coast-town of Veile (or Vejle), two barrows, locally known as the graves of King Gorm and his queen, stand by the churchyard. Hard by are ancient stone monuments, bearing runic inscriptions[96].

Nor do these Northern cases lack counterparts elsewhere. The church at Arrichinaga, in the province of Biscay, in Spain, was so built as to enclose the huge stones of a great dolmen; between the stones is placed the shrine of the patron saint[97]. The rugged land of Brittany is well-known to all travellers for its illustrations of lingering paganism; to some of these we shall again refer. But if we desire to learn how imperative, how inescapable, was the spirit of compromise, we should turn to the works of old writers, such as that curious old volume which relates Jean Scheffer’s travels in Lapland in the latter part of the seventeenth century[98]. There, we shall discover a strange alloy of heathenism and Christianity, visible to all, seemingly condemned by none. Even in our own day, so recently as the year 1895, we hear of curious practices among the Samoyads. These folk, though nominally Christians, within modern times still sacrificed human beings clandestinely, and conducted heathen services within the ancient stone-circles, carefully screening the images of their gods from the public gaze.

Returning to the high road of our inquiry, we ask whether these examples can be paralleled in Britain. Consider for a moment the great wealth of our folk-lore, our superstitions, our almost incredible heathen practices. Grease from the church-bell to cure rheumatism; pellitory from the church-wall for whooping-cough; teeth from the graveyard to serve as charms; the midnight watchings on St Mark’s Eve; the folk-tales about evergreens; the superstitions connected with baptisms, marriages, and deaths; the hundred and one little beliefs which run in an undercurrent beneath the apparently smooth surface of religious thought—do not these suggest that we may expect to find parallels to the continental examples of church-building on heathen soil? How strange if our islands had escaped the influences which are seen in almost every other European country! Yet, to speak plainly, our direct testimony is very scanty.

We know that, at Rome, the Pantheon became a Christian church, and we have previously mooted the possibility of pagan idol-temples having been similarly treated in Britain. Conclusive proof cannot be given, since subsequent restorations would erase, or at least obscure, the vestiges which we seek. Professor Baldwin Brown admits two possible examples, without committing himself to a decided opinion. One is the church of Silchester previously noted (p. 23 supra), and the other that of St Martin’s, Leicester. The latter church rests on the site of a Roman columnar structure, which would have been suitable for a temple[99]. There are also certain clues afforded by tradition and philology. At Woodcuts Common, in Cranborne Chase, there is an imperfect amphitheatre known as Church Barrow, which was excavated by General Pitt-Rivers. This high authority suggested that the depression which forms the arena was used for games, and, not improbably, in early Saxon times, before any church was built in the neighbourhood, for divine worship[100]. Mr Allcroft gives reasons for supposing that the present earthwork is on the site once occupied by a tumulus[101]. Whichever hypothesis be accepted, the name of Church Barrow will not be lightly set aside by the folk-lorist, for it does not stand alone. At a spot called Church Bottom, or Sunken Kirk, near Ickleton, in Cambridgeshire, Roman relics, suggestive of a columnar building, were discovered. Pitt-Rivers supposed that a Roman basilica, for Christian worship, existed on the site, and that it was re-adapted when the East Anglians became converted to Christianity[102]. The data, in this instance, are not plentiful, and one might perhaps conjecture, with equal reason, that the original building was pagan. An earthwork on Temple Downs, a few miles north of Avebury, Wiltshire, was traditionally called “Old Chapel.” By the way, we notice that the names of Kirk, Old Kirk, Sunken Kirk, and Chapel Field, as applied to earthworks and sites containing ancient foundations, are not uncommon[103], and one is naturally led to connect this fact with the known association of churches and earthworks. Again, at Llangenydd, Glamorganshire, there may be seen, in a field, the remains of a stone-circle which is still called Yr Hen Eglwys, “the old church,” the meadow being known as Cae’r Hen Eglwys, “Old Church Field.” Tradition says that here the inhabitants worshipped before the present church at Lalestone was erected. A remarkable parallel is exhibited in the Shetlands, where churches were often built, we are assured, amid the ruins of heathen “temples.” The analogy consists in this: that the word “kirk” is now applied to holy spots, whether a chapel exists there or not. Again, Sandwich Kirk, in the island of Unst, represents the ruins of a reputed chapel which stood beside an ancient kitchen-midden. At Kirkamool, bones and pottery were dug up under the foundations of the sacred building[104]. Germane to this subject, one may mention the old ruins of Constantine Church, in Cornwall, which lie near an old kitchen-midden, and which have yielded to the spade of the explorer bones of men and domestic animals, besides pottery of the Mediaeval, Roman, and Neolithic periods[105].

Pursuing the trail provided by philology, one must glance cursorily at the theory propounded by Isaac Taylor that place-names like Godshill, Godstone, Godley, Godney, Godstow, and Godmanchester, are mute witnesses of the substitution of the new faith for the old[106]. The theory is certainly plausible, but, as Professor W. W. Skeat pointed out in a letter to the present writer, the question can only be settled by an appeal to carefully compiled name-lists, especially those which give the spellings that were current during the Middle English period. Now it chances that my friend Mr A. Bonner has, for many years past, been making researches on these, and similar place-names, and he has kindly allowed me the use of his unpublished work. Thus, by the aid of Professor Skeat and Mr Bonner, one is able to test the theory that these particular names commemorate the establishment of Christian worship. To begin with, it must be observed that, owing to the modern defective and misleading system of orthography, not only may origins be disguised, but one mode of spelling may hide several possible etymologies. Thus, the A.S. god (=good) is frequently confused with god (=God); moreover, since God (=Good) was also a personal name in Anglo-Saxon, we may get further complexity; e.g. Goodrich (A.S. God-ric), in Herefordshire. Dealing with a few of the names mentioned above, we have Godstone, Surrey, appearing in the thirteenth century as Codeston and Coddestone, the spelling God being of much later date. Though the question cannot be settled in the absence of an Anglo-Saxon form, it is probable that the word denotes a personal name. Godney, Somerset, apparently represents “Goda’s island,” and Godley, the name of a hundred in Surrey, “Goda’s meadow.” Godshill (Gods, i.e. Good’s hill), Godstow, and Godstoke, again, all give indications of personal names.

Goodmanham, or Godmundingham, near Market Weighton, in Yorkshire, is believed to be the spot mentioned by Bede as having a celebrated pagan temple, and as being the scene of missionary work by Paulinus. Isaac Taylor was at first content to follow Grimm in deriving the word from the Norse (godi =priest, and mund, protection of the gods). Afterwards he discovered his error, for, in a later edition of Names and their Histories, he explains the name as the “home of the Godmundings or descendants of Godmund.” Alas, in a posthumous edition, very recent, of Words and Places, the old blunder creeps in again. Mr Bonner states that the earliest form (c. A.D. 737), is Godmunddingham; the tenth century spelling was practically the same; hence the meaning is, “ham of the Godmund or Goodman family.”

Again, the village-name of Malden, in Surrey, has been claimed as meaning “Hill of the Cross,” and as indicating the turn-over to Christianity. It is true that the Anglo-Saxon mÆl means a mark, and that the Domesday form, Meldone, points to a down on which stood some mark, probably a beacon or boundary post. Yet, although the village church is situated on the highest spot in the neighbourhood, whence the ground slopes away on two sides, the building does not stand on the Chalk downland, but on the London Clay. Moreover, although we have many post-Conquest orthographies, such as Maudon, Meaudon, and Maldene, to guide us, we lack evidence concerning the true A. S. form. Had the name been Christ’s Maldon (Cristes mÆl dun), it would certainly have implied a “hill with a cross or crucifix,” just as, according to Professor Skeat, we have “Christ’s mÆl ford,” now oddly turned into Christian Malford (Wiltshire). The evidence for Maldon, in Essex, is more satisfactory than that for the Surrey village. Here we get the tenth century spelling MÆldune, and, although the modern town is built on low ground by the river Blackwater, a hill, surmounted by a tumulus, rises behind to a height of 109 feet. Hence this name may perhaps be the equivalent of “Hill of the Cross.”

The prefix Llan, which occurs so frequently in Wales and the Marches, affords a surer indication of a period when the possession of a church by the village community was the exception, and not the rule. Not only this; but expert opinion shows that llan signifies an enclosed or fenced-in space. The reference is therefore rather to the churchyard than to the sacred fabric, and it is believed by some that the word retains memories of the worship held within stone-circles. When we come to consider the relation of Bardic assemblies with parish churches, it will appear that this supposition is reasonable. Even in Wales, the prefix Llan-is sometimes replaced by Kil-, as in Kilfowyr and Kilsant, in Caermathenshire; but it is chiefly in Scotland that we look for such place-names. The word kil originally meant a hermit’s cell, and afterwards came to be applied to a church. Finally, as an aid in detecting places which possessed churches at an early date, and which were thus pre-eminently worthy of special designations, we have the Norse, and Danish, prefix, Kirk-, as in Kirby, Kirk Ella, Kirkcolm. Premising that a little additional testimony under the head of philology will be given later, we must now follow another clue.

On the whole, it must be conceded that the support derived from geographical names is somewhat feeble, yet it may prove capable of being extended as knowledge increases. Our next line of research gives fairer promise; it brings us to examine the ancient rude monuments which are frequently found in the vicinity of village churches.

The megalithic monuments recognized by the archaeologist are of several kinds, but we shall be here concerned with three of these groups—the menhir, or single upright stone; the cromlech, or stone-circle; and the dolmen, or “stone-table.” These prehistoric remains seem to have seriously attracted the notice of the Teutonic invaders, who were prone to follow idolatrous practices based upon lingering traditions about the storm-fretted stones. To this superstitious respect attention has been drawn in a previous work[107]. Some indications of the honour imputed to these megaliths is gleaned from a study of parish boundaries, though it is almost certain that many of the stones erected in such positions belong to the historic period. The old open-air tribunals, too, were wont to meet at barrows, cairns, cromlechs, and menhirs, and at the foot of the crosses by which the menhirs were largely supplanted. This statement holds true for other places besides England. In the churchyard of Ste Marie du Castel, Guernsey, there existed three large stones, which marked the spot where open-air courts were held until recent years[108]. Evidence is also obtainable from several countries, showing that the election and coronation of kings and princes were associated with stone-circles. Nor, indeed, were our ancestors very exigent in this matter; a rude natural boulder or monolith was considered a good substitute for the artificial pillar which had been erected by forgotten folk. Over and over again we meet with “blue-stones”—chiefly glacial boulders—which were set up to mark the limits of a parish, or to form the trysting-place of a manorial court[109]. Lastly, it is on record that Patrick, Bishop of the Hebrides, desired OrlygÜs to build a church wherever he found the upright stones or menhirs.

Fig. 8. The Agglestone, Studland Heath, Dorsetshire. A natural mass of concretionary sandstone belonging to the Bagshot sands of the district. Much pagan tradition is associated with this block, which has been curiously eroded by rain, frost, and wind. The so-called “Druid’s basins” are altogether natural cavities.