They only write the get with something whose mark lasts, like ink or red paint or Komos or Kankantom (explanation: types of pigment) and the like; but if one wrote it with something that does not last, like beverages or fruit juice or the like, it is not a [valid] get.
If he wrote [the bill of divorce] post facto on Avar, Shechor, or Shichur, it is valid. (The meaning of Avar is lead, Shechor is an iron utensil used for writing, and Shichor is prepared coal.) Rem"a: There are those who say that lead is valid only when pieces of lead are mixed into water (Tur and Ro"sh), but if he wrote with lead itself, it is invalid, and one should be concerned about this. Preferably, the practice is to write with good ink like that used for writing books [of the Bible] (Seder Gitin), and this should not be changed. There are those who are even more stringent and say that the scribe must read each letter before he writes it, as with books [of the Bible] (ibid.).
We write it (the get) with gallnut liquid, as long as the parchment was not itself tanned with gallnut, for gallnut liquid is not legible on such [a parchment]. And if he did write [with gallnut liquid on gallnut tanned parchment] the get is not valid. We apply the same reasoning with other similar cases [of combinations of ink and parchment].
If the bill of divorce was inscribed on a board, stone, or metal tray, if the legs of the letters have been dug in, it is valid. So too if he engraved the legs of the letters from behind the tray so that they protruded from the front of the tray. However, if he dug into the letter until the legs appeared at different heights, or he hit the front of the tray until the letters protruded like on a gold coin, it is not a bill of divorce. Rem"a: Some say that preferably one should be stringent and not write with an iron quill, so that he should not come to emboss (Seder Gitin). See below, clause 22. It is not considered to be embossed unless the entire letter is formed that way, but if [he partially embosses a letter that] still needs to be fixed so that it may be a letter, it is valid (Terumat HaDeshen 228).
If he cut out of the leather the shape of the writing, it is valid. But if embossed the writing into the leather, it is invalid. Some authorities hold the opposite, and there is one authority who rules both cases valid.
If the practice was to be stringent regarding letters being surrounded by parchment [i.e., empty space], and there were letters stuck to one another, he must scratch between them to separate them. Rem"a: Some say that one should not separate letters that are touching (Seder Gitin), and this is our practice. There are those who are lenient in pressing time (Responsa of the Ro"sh) who allow fixing before the witnesses sign (Mahara"m Padua), when the husband commands that it be fixed (Beit Yosef on the view of the Ro"sh).
If someone wants to thicken the letters, we protest against him.
If a drop of ink fell within the letter (Tur) and it is unrecognizable, he may not remove the ink and fix the letter, as that would be considered embossing. There are those who are stringent even if the letter is recognizable (Mahari"k 98), and this is our practice.
If he wanted to form an open Mem [i.e., in a medial position], but his hand continued and he closed it [and it became a final Mem], he may not scratch away that which is stuck together, as that is considered engraving. (See Orach Chaim 32.)
The custom is to etch lines [on the parchment]. However, one should avoid etching with lead for it colors, and the outcome will be text upon text.
Some have a custom to write it [the get] on 12 lines.
It is customary to make a 13th line and divide it in two, and on that line the witnesses sign one underneath the other. Rama: and they also etch two small lines [for the witnesses]. He should etch the lines before he starts to write the get. He should etch from the sides, and the lines of the get should be equal, not one long and one short. If he did not etch lines at all, it is valid. Likewise, if he made more or less than 12 lines [it is valid] (this is all in Seder Gittin, also in Mordechai beginning of Tractate Gittin). Some say that we should not validate them (gets with atypical number of lines) except where we are worried about "anchoring" (leaving the woman unable to remarry), that the agent arrived from a distant land and caravans are uncommon. It seems to me that the same rule applies to etching lines. He must start the etched lines outside the text (Beit Yosef in the name of Kuntres and in Seder Gittin). He should write the get on the flesh side of the parchment and not the skin side. If he wrote it on the skin side, some invalidate the get (Maharil). It seems to me that if there is a fear of anchoring, we should not be stringent. When the scribe writes, it is preferable that he should be careful not to touch (i.e. to write) the etching at the sides. It is customary ex-ante to leave a space of 1/2 finger on the sides, and one whole finger on the top (Seder Gittin), and at the bottom space to grasp with the hand (Maharil) or more (Rama's own opinion).
It is customary to make the get such that its length... Rama: including the blank space at the bottom (Bet Yosef in the name of Kuntres) Mechaber: ...is greater than its width.
It was the custom that the letters should not go outside the guidelines on either the right or left side. (Rema: If they were written outside the guidelines, there are those who say even one letter will invalidate it, but in a time of need there are those who are lenient.)
There should be no erasure, and they should not write it on an erasure. Rama: If a fly landed and dragged the ink, he should not erase the marks (Seder Gittin). He should not write on any hole over which the ink does not pass over (ibid.).
The text should not be twisted or contorted such that one letter can be confused with another and as a result the meaning will change. Rama: If the feet of the heh and the kuf touch together, it is invalid even ex post facto (Terumat haDeshen chapter 236 and in his rulings chapter 45). However, if a minor can read the text, some authorities still validate when there is a concern for anchoring (putting her in limbo of not married, but not able to remarry) the woman (Maharik root 71). At the outset it is forbidden to separate [the feet of the letters] with a knife, but he should write another get (ibid. root 98). But only where there is no concern for anchoring, but in times of oppression in which there is a concern for anchoring, he may fix the print (opinion of Rama), and see Orach Chaim chapter 32 how to correct. See chapter 126 se'if 48. If he needs to fix letters that do not touch in places where they should touch, or they are short and need to be lengthened, he should fix them before he writes (at the end of the get) the words "k'dat Moshe v'Yisrael" (Hagahot Seder Gittin, a tradition from Mahariv, and so I also found written in the name of Maharam Meor Einayim); or he should at least fix the letters before the witnesses sign (Maharam Padua). He should be careful that everything that needs to touch does touch, and things that should not touch do not touch, just as in writing a scroll of Torah, tefillin, and mezzuzot (Seder Gittin, and so implied in Hagahot Mordechai Gittin). If letters that should be touching are not touching, such as the yud of the tzadi (facing rightward) or the yud of the aleph, and other similar letters, some authorities say that even if a young child who is neither wise nor stupid can still read it, is is nevertheless invalid even ex post facto (Terumat haDeshen and in his rulings chapter 43). Some authorities say that where there is a concern for anchoring the woman, it is appropriate to validate in such a case. If the error is in the "tofes" (form text of the get except for names, date, place) it is appropriate to be lenient even when there is no concern for anchoring (Beit Yosef). If he wrote the lamed and aleph of Yisrael together as a single letter (this is a common contraction), some say that it is valid if it was in the tofes of the get (Responsa Rosh), and where there is a concern for anchoring, it is appropriate to be lenient. If the anterior point of the yud on the aleph also touches in front, it is still valid (found written in the name of Mahari Maribor = R. Israel Isserlein, author of Terumat haDeshen). If the letters spread out because of wateriness of the ink that is called "Giplasen", so that it appears that the letter may be touching but it can be seen that they are not actually touching, it is valid (found written in the name of Mahari Bruna =Brno in the Czech Republic).
If the text could be interpreted in two different ways, or the if writing was distorted or confused, or it is possible to read it in a different manner, it is invalid.
The get must be readable to a young child, neither smart nor stupid, in the locale of its delivery. There is an authority who says that it must also be readable in the locale in which it was written. Rama: [When checking readability] it is not necessary to cover around the sides of the letter in question, rather they show him the document as is, and if he can read it, it is kosher (found written in an old arrangement). It is customary to mix (or pour) the lines, which is to extend the inferior extensions of the nun sofi and peh sofi from the upper line until the line below, or the head of the lamed from the line below into the line above. If he did not intermix the lines, some say that it is invalid, but to me it appears that if there is a concern for anchoring the woman, it is not appropriate to be stringent. He should be careful not to extend the head of a letter from a lower line into the inner space of a letter from the line above or vice versa (Ran chapter 2 of Gittin, in the name of "some say"), and if he did this and the shape of the letter has been changed, for example he extended the head of a lamed into the space of a dalet, and the dalet now looks like a heh, then it is invalid. (Opinion of the Rav (Rama) specifically in similar cases, and so implied in Hagahot Semak).
A get on which a letter or word has been erased, or written between the lines, if this is part of the tofes of the get, it is valid. But if it is part of the toref, it is not a get. And if he goes back and explains at the end of the get that a certain letter was written between the lines, or over an erased part, even if this is part of the toref of the get, it is valid, as is true for other documents. And the upholding of the get should be written before "as is the law of Moses and Israel." And now we have the custom of being strict and not upholding a get that had been erased.
A get which is torn vertically and horizontally, which is how it is torn in court, is an abrogated get as is any other document [torn in similar manner]. But if it is torn in a manner different from the way it is torn in court, it is valid. If it had dissolved or rotted or become sieve-like, it is valid. If it was erased or smudged, but its shadow is still present, if it is readable, it is valid, but if is not readable, it is not a get.
What are the circumstances? When the get is in her possession with signed witnesses, and there are no witnesses to the delivery of the get. However, if there are witnesses before whom the get was delivered and it was kosher at the time, the get is valid even if the toref (names, date, place, signatures, as opposed to the text) was written on an erasure or between the lines, or it was torn vertically and horizontally when it was given to her before the witnesses. There is an authority when it was torn vertically and horizontally, Rama: and likewise when it was smudged to the point that it is not readable (so wrote the Magid Mishne).
There is one authority who says that one should be careful not to write the get with a pen made from a wing, but rather from a reed. See infra se'if 4.
If he embroidered or wove or made needlework in the shape of the letters of the get, some say that this is not considered writing (and the get would be invalid).