That is to say, the Lord of Richard's castle, on whose surrender and punishment Godwine had specially insisted, was specially exempted, as guiltless, when Godwine returned to power.[13]
In me, at least, this discrepancy aroused grave suspicion, and I turned to see what foundation there was for identifying the offending garrison of 1051 with that of Richard's castle. I at once discovered there was none whatever.
We have here, in short, one of those cases, characteristic, as I think, of the late Professor's work, in which he first formed an idea, and then, under its spell, fitted the facts to it without question. The view, for instance, of the unique position of Richard's castle as 'the castle' at the time is at once rendered untenable by the fact that, on the return of Godwine, Normans fled 'some west to Pentecostes castle, some north to Robert's castle', in the words of the Chronicle.[14] Moreover, the former belonged to Osbern, 'whose surname was Pentecost' (cognomento Pentecost), who, as we learn from Florence, was forced to surrender it and leave the country, as was also the fate of another castellan, his comrade Hugh.[15]
It is important to observe the clear distinction between Richard, son of Scrob, of Richard's castle, and Osbern Pentecost, of Pentecost's castle, of whom the former was allowed to remain, while the latter was exiled. But it is another peculiarity of Mr Freeman's work that he was apt to confuse different individuals bearing the same name.[16] In this instance, he boldly assumed that 'Pentecost, as we gather from Florence [?] ... is the same as Osbern, the son of Richard of Richard's castle, of whom we have already heard so much' (ii. 329), although the latter, a well-known man, is always distinguished as a son of his father, and never as Pentecost. And he further assumes that 'Pentecost's castle' was identical with Richard's castle, 'the first cause of so much evil' (Ibid.). These identifications led him into further difficulty, because Osbern, the son of Richard, is found afterwards holding 'both lands and offices in Herefordshire' (ii. 345). To account for this, he further assumes as 'certain that Osbern afterwards returned' (Ibid.). This assumption led him on to suggest that others also returned from exile, and that 'their restoration was owing to special entreaties of the King after the death of Godwine' (ii. 346). The whole of this history is sheer assumption, based on confusion alone.
Now let us clear our minds of this confusion, and keep the two castellans and their respective castles apart. On the one hand, we have Richard, the son of Scrob, who was left undisturbed at his castle, and was succeeded there by his son Osbern;[17] on the other hand, we have Osbern, 'whose surname was Pentecost', and who had to surrender his castle, to which the guilty Normans had fled, and to go into exile. Can we identify that castle? I would venture to suggest that it was no other than that of Ewyas Harold in the south-west corner of Herefordshire, of which Domesday tells us that Earl William had re-fortified it ('hoc castellum refirmaverat'), implying that it had existed, and been dismantled before the Conquest. It heads, in the great survey, the possessions of Alfred of Marlborough, and although its holder T.R.E. is not mentioned, we read of the two Manors which follow it: 'Hæc duo maneria tenuit Osbernus avunculus Alveredi T.R.E. quando Goduinus et Heraldus erant exulati' (i. 186). Mr Freeman, of course, assumed that this Osbern was identical with Osbern, the son of Richard, the Domesday tenant-in-chief. This assumption is not only baseless, but also most improbable: for Alfred was old enough to be father-in-law to Thurstan (Mortimer), a Domesday tenant, and would scarcely therefore be young enough to be nephew to another Domesday tenant-in-chief. I would suggest that his uncle was that Osbern 'Pentecost' who had to surrender his castle and flee on the return of Godwine and Harold. This would exactly fit in with the Domesday statement, as also with the dismantling of Ewyas Castle.[18]
Ewyas Harold fits in also with the chronicle's mention of the Normans fleeing 'west' to Pentecost's castle.
We have now seen that Richard's castle did not stand alone, and that there is nothing to identify it with that Herefordshire castle ('ænne castel') of which the garrison had committed outrages in 1051, and which is far more likely, so far as our evidence goes, to have been 'Pentecost's Castle'. Mr Freeman rightly called attention to 'the firm root which the Normans had taken in Herefordshire before 1051, which looks very much as if they had been specially favoured in these parts' (ii. 562); and he argued from this that Earl Ralf had probably ruled the shire between 1046 and 1050. The Earl would naturally have introduced the foreign system of castles, as he did the foreign fashion of fighting on horseback. Indeed, speaking of the capture of Hereford in 1055, Mr Freeman wrote:
It is an obvious conjecture that the fortress destroyed by Gruffyd was a Norman castle raised by Ralph. A chief who was so anxious to make his people conform to Norman ways of fighting would hardly lag behind his neighbour at Richard's castle. He would be among the first at once to provide himself with a dwelling-place and his capital with a defence according to the latest continental patterns (ii. 391).