Siman 356

We learned in a Mishnah in Eruvin (87a): "A water channel that passes through a courtyard—one may not draw water from it on Shabbat unless they made for it a partition ten tefachim high at the entrance and at the exit." And this is its explanation: that the water channel extends from afar, and passes through the courtyard from this side and exits through the other side and extends further. Therefore, if it is only ten deep and four wide—it is not nullified to the courtyard, like all things in the courtyard whose law is as a reshut hayachid as was written in siman 345, for that is only if it had no connection to the outside. Therefore, its law is as a karmelit, and it is forbidden to carry and to draw water from it to the courtyard and from the courtyard into it. However, if it is not ten high or is not four wide—it is not significant, even though it is connected to the outside, and it is nullified to the courtyard, and its law is as a reshut hayachid. And so too with the rectification that will be explained, it separates that which is in the courtyard from that which is outside of it.

And the remedy is—to make for it a partition of ten tefachim at the entrance and at the exit, which is to say either nine tefachim outside the water and one tefach inside the water, or nine inside the water and one tefach above the water, in order to differentiate between water and water. And even though in the previous siman regarding a balcony, we did not require even that the partition be near the water, and certainly not inside the water, and why do we require here also a bit of the partition inside the water? It is not comparable, for we already explained there that the permit of a suspended partition in water requires that there be a recognition that it was made for the purpose of the permit of a partition, and therefore the partitions of a latrine do not avail there, see there. And according to this it sits well, for regarding a balcony where the partitions are around the hole of the balcony, through which they draw the water, everyone sees that the partition was made for the purpose of the permit, and therefore they do not need to reach the water, and certainly not inside the water, for regarding the water itself partitions will not avail, as they mix from below. This is not the case regarding a water channel that passes through a courtyard; if we make a partition above the water, there will be no recognition, for behold these are the partitions of the courtyard as in all courtyards. Therefore they required that there be a bit of the partition inside the water, for through this it is recognizable that it was made for the purpose of the permit. (I learned this from the words of Rashi there, s.v. "Asu," see there).

And know, that in this law there is no difference whether the water channel enters the courtyard with a width of ten amot or less, and whether it enters in its entirety, meaning that no partial partitions of the courtyard remained from the side, or whether it did not enter in its entirety, such that partial partitions remained and with less than ten, and nevertheless it is forbidden to fill from it to the courtyard. And one should not ask why it will be explained in this siman, regarding an inlet of the sea that enters a courtyard, that it is not forbidden to fill from it to the courtyard unless the breach is more than ten, or it is in its entirety, but otherwise it is permitted to fill from it to the courtyard—it is not comparable. For there it is breached only from one side of the courtyard, and its law is like all breaches, where less than ten and remaining side-posts have the law of an opening and it does not forbid. This is not the case with a water channel that extends through the courtyard from both sides, and it is recognizable to all that it extends from a distance from both sides, and it is considered a domain unto itself, and we cannot judge it with the law of an opening, since it is perforated from side to side. Therefore, even with less than ten and it is not in its entirety—it is forbidden to draw from it to the courtyard. (I learned this from the words of the Ritva) And nevertheless, when it is not ten deep or is not four wide—it is not significant, and it is nullified to the courtyard as I have written. (ibid.)

If the water channel enters the courtyard with a width of less than three tefachim—it requires no rectification, (Ritva) for anything less than three is considered lavud. And even if it enters in several places, and in each place it is less than three—it is nothing. Therefore, even if one began to make the partitions from each side and did not complete them, in that he did not connect them in the middle so that the water would have a path to enter and exit through there, if there is not three tefachim between them—it is permitted, and if there is three tefachim—it is forbidden, and similarly ab initio, it forbids with more than three tefachim. And even though less than four tefachim is not a significant place, nevertheless, if the channel itself is four wide and ten deep, the breach forbids even with less than four. (And in the Taz, subsection 4, he wrote that with a width of four it requires partitions, and it is astonishing, and perhaps it is a printing error and it should say three, and so wrote the Pri Megadim).

And know, that all this is regarding the matter of carrying the water from the water channel. However, that the water channel should prohibit carrying in the entire courtyard—it does not prohibit. And it is not only necessary to say in a case of less than ten where the side-posts remained, but even if it was breached more than ten, or if it was breached in its entirety—it does not prohibit, for since the water channel is ten deep and four wide as has been written, it follows that the water channel itself becomes a partition. (Magen Avraham, subsection 2 in the name of the Ra'avad) And even if at its beginning it is not ten deep, if it rises ten within four—it is a partition, but if it is not so—it prohibits when it is more than ten, or in its entirety. (ibid.)

The Tur and the Shulchan Aruch wrote in section 2: "A courtyard that was breached, and an arm of the sea passes over the breach, if it is not in its fullness and there is not more than ten in the breach—it is permitted to draw from it and to bring it into the house. If it was breached in its fullness, or if there is more than ten in the breach, if there remained in the place that was breached banks ten high and the water covers them—it is permitted to draw from it into the courtyard, but it is forbidden to bring them into the house, unless they made a partition ten high over the water. If banks did not remain—it is forbidden even to draw into the courtyard." Thus far their words.

The explanation of the matter is that when a courtyard is breached on one side and an arm of the sea enters it, if it is less than ten and not in its full width, it is like an opening, and the water that entered is like a pit in the courtyard, and the partitions of the courtyard separate it from the waters of the sea that are before the courtyard; it is not similar to a water channel that is breached from two sides, as has been written. However, if the breach is more than ten or in its full width, it prohibits the courtyard entirely, even to carry within it, because it is breached into a prohibited area. And if there remained banks of ten that are visible, it is permitted entirely, both to carry in the courtyard and to draw from the water to bring them into the house, since there is a complete partition. However, if these banks are not visible and the water passes over them, to fill from the water in the courtyard is permitted, for they are a partition even though they are covered, because sometimes the water will subside and the partitions will be seen. But to bring them into the house is prohibited, for in the house the partitions are not recognizable at all. Nevertheless, if the arm is ten deep, it does not prohibit carrying in the courtyard, for it itself is considered like a partition, as has been written in section 5 regarding the channel, but it is prohibited to fill from it, since there is no cessation of a partition from above to separate between it and the water outside, for it was breached in its full width or more than ten, and the sea is a karmelit. And one should not ask: even when there are banks, why do we not fear lest the sea cast up a sandbank, as we fear in siman 363, see there? For it can be said that it deals with a case where the banks are very high, or it is very far from the seashore. (Tosafot Eruvin 12a, starting with "Haka") And all this is the system of Rashi, Tosafot, and the Ra'avad in chapter 15, and the Tur and the Shulchan Aruch. And according to the Rambam there, he has another system in this, see there.