Keilim, Chapter 28, Mishnah 6. Our mishnah continues the chapter's discussion of a matlit, a patch of cloth, and asks what happens when a patch that already carries tum'at midras is sewn onto another object. Everything turns on one distinction: is the object it was attached to something that is itself capable of receiving tum'at midras, or not? A basket is the example of an item that cannot receive midras; a garment is the example of an item that can.
Before entering the cases, recall the terms. An item pressed upon by a zav or zavah becomes a midras, and its status is that of an av hatum'ah, a primary source of tum'ah. Whatever an av hatum'ah touches becomes a rishon, first level; whatever a rishon touches becomes a sheni, second level; and a sheni that touches terumah does not make it tamei but renders it pasul, invalid. That is why the mishnah speaks of an object that "transmits tum'ah to so many levels and invalidates one level."
The patch sewn onto a basket
The first case is introduced with the words "Matlit shetelayah al hakupah": a patch already bearing tum'at midras that has now been fastened onto a basket. A basket does not belong to the category of items able to take on midras, so the attachment cannot pass that severe status along. What does pass along is ordinary tum'ah by contact: the basket came into contact with an av hatum'ah, and it therefore holds first-level status, a rishon. Hence "metame'ah echad u'foselet echad": one level of tum'ah is conveyed and one level is invalidated. Food that touches this basket becomes a sheni, and should that food then reach terumah, the terumah is disqualified.
The mishnah continues, "Hifrisha min hakupah": the patch is afterwards taken off the basket. The basket does not shed what it already picked up, and of it the mishnah says that it is "metame'ah echad u'foselet echad", still a rishon that conveys one level and disqualifies one. Of the patch, however, the ruling is "vehamatlit tehorah", it emerges pure. From the moment it was stitched on it counted as part of the basket, and since a basket is incapable of bearing tum'at midras, the patch forfeited its midras status when it merged into it. Detaching it is no different from breaking a piece off the basket itself: nothing of its former tum'ah returns.
The patch sewn onto a garment
Now the mishnah turns to a patch attached to something that can receive tum'at midras, such as another piece of cloth. Here the patch does not lose anything by being joined, because the garment is a fit candidate for midras. And while the two are joined, the garment shares the patch's severe status: this is tum'ah bechiburin, tum'ah through attachment, so that as long as they are connected the garment too counts as an av hatum'ah.
"Telayah al habeged, metame'ah shenayim u'foselet echad." Since the garment now has the standing of an av hatum'ah, it transmits tum'ah through two levels and invalidates one. Food touching the garment becomes a rishon; food touched by that rishon becomes a sheni; and when that sheni reaches terumah, the terumah is invalidated.
"Hifrisha min habeged": the patch is then removed from the garment. The garment loses its standing as an av hatum'ah, yet it does not return to purity. On account of the contact it had with the patch it remains at first level, and so the mishnah rules of it "metame'ah echad u'foseil echad": as a rishon it renders food a sheni, and that sheni goes on to disqualify terumah, one level transmitted and one level invalidated.
Of the patch the mishnah says "vehamatlit metame'ah shenayim u'foselet echad." Separation costs it nothing: it carries the rank of av hatum'ah exactly as before, conveying tum'ah through two levels and disqualifying at the third. Whatever touches it becomes a rishon, that rishon produces a sheni, and the sheni disqualifies terumah. Its severity was never interrupted at any stage, not before it was stitched on, not while the two were joined (since the cloth to which it was fastened is itself fit to receive midras), and not after it was taken off.
Sack and leather: three opinions
Rebbi Meir extends these rulings further: "Vechen hatoleh al hasak", and likewise "o al ha'or", a patch hung on sackcloth or on leather, "divrei Rebbi Meir". Sack and leather are themselves materials subject to tum'at midras, and in his opinion the difference in material between the cloth patch and the object it is sewn to makes no difference; the patch keeps its tum'at midras when joined to a fabric of another kind, and everything taught above applies.
"Rebbi Shimon metaher." According to Rebbi Shimon, once the patch has been sewn onto sack or onto leather it is cleansed of its tum'at midras. Cloth and sack, or cloth and leather, are not the same species of material, and that dissimilarity causes the patch to lose its midras status upon being attached.
A third view is recorded in the name of Rebbi Yose, who divides between the two materials. His words are "al ha'or tahor": attached to leather, the patch comes out pure, for cloth bears no resemblance to leather and the patch is never genuinely absorbed into it. But "al hasak tamei": on sackcloth it holds on to its tum'at midras, and the reason given is "mipenei shehu arig", because sack is woven. Cloth and sack are relatives of a sort, both being products of weaving, and therefore the patch does not lose its midras when it is joined to sack.
The mishnah thus teaches a single principle applied through several cases: a patch bearing tum'at midras keeps that severe status only when it joins an object that could bear such tum'ah itself, and even then the Tannaim differ over how much similarity of material is required for the joining to count.