Even against his lord the bondman had some standing ground for a civil action. It has rightly been maintained, that he could implead his master in consequence of an agreement with him. The assertion is not quite easy to prove however, and has been put forward too sweepingly[86]. At first sight it seems even that the old law books, i.e. those of Bracton and his followers, teach the opposite doctrine. They deal almost exclusively with the case of a feoffment made by the lord to a villain and his heirs, and give the feoffee an action only on the ground of implied manumission. The feoffor enfranchises his serf indirectly, even if he does not say so in as many words, because he has spoken of the feoffee's heirs, and the villain has no other heirs besides the lord[87]. The action eventually proceeds in this case, because it is brought not by a serf but by a freed man. One difficult passage in Bracton points another way; it is printed in a foot-note[88]. There can be no doubt, that in it Bracton is speaking of a covenant made by the lord not with a free man or a freed man, but with a villain. This comes out strongly when it is said, that the lord, and not the villain, has the assize against intruders, and when the author puts the main question--is the feoffor bound to hold the covenant or not? The whole drift of the quotation can be understood only on the fundamental assumption that we have lord and villain before us. But there are four words which militate against this obvious explanation; the words '_sibi et heredibus suis_.' We know what their meaning is--they imply enfranchisement and a freehold estate of inheritance. They involve a hopeless contradiction to the doctrine previously stated, a doctrine which might be further supported by references to Britton, Fleta and Bracton himself[89]. In short, if we accept them, we can hardly get out of confusion. Were our text of Bracton much more definitely and satisfactorily settled than it is[90], one would still feel tempted to strike them out; as it is we have a text studded with interpolations and errors, and it seems quite certain that 'sibi et heredibus suis' has got into it simply because the compositor of Tottell's edition repeated it from the conclusion of the sentence immediately preceding, and so mixed up two cases, which were to be distinguished by this very qualification. The four words are missing in all the MSS. of the British Museum, the Bodleian and the Cambridge University Library[91]. I have no doubt that further verification will only confirm my opinion. On my assumption Bracton clearly distinguishes between two possibilities. In one case the deed simply binds the lord as to a particular person, in the other it binds him in perpetuity; and in this latter case, as there ought not to be any heirs of a bondman but the lord, bondage is annihilated by the deed. It is not annihilated when one person is granted a certain privilege as to a particular piece of land, and in every other respect the grantee and all his descendants remain unfree[92]:--he has no freehold, but he has a special covenant to fall back upon. This seems to lie at the root of what Bracton calls privileged villainage by covenant as distinguished from villain socage[93].