Future Fixes for JSesh fonts

Dr. Peter Dils, from Leipzig, sent me (a long time ago) a list of signs for which the Hieroglyphica had significantly different shapes from the signs in JSesh.

Some of those differences hint about problems in the way we encode hieroglyphs - basically, upon what should be encoded or not. But others are really error in the JSesh shape, and should be fixed.

I have extracted from the list those signs which, in my judgment, are in most urgent need of redrawing.

Note: this list on based on JSesh version 7.6.1. A number of those problems will be fixed in the next JSesh version.

1. Non problematic signs

A sign can be modified in a font without any problem if the change does not change the identity of the sign and its value. Basically, if it doesn't break the encoding of existing texts.

For instance, the shape of M3 is in JSesh. When I started creating the original font, my idea was that I would use facsimile from actual ramesside texts. The problem is that, in the current fonts, you want to choose a typical shape, not just any existing shape. At that time, I was mostly interested in making linguistic databases, and the idea of covering the whole extended library was a distant dream. If you don't have many variants of the same sign, then you can choose one of them more or less freely, which is what I did. But, now, we have added some of the variants from the Hieroglyphica to the font, so the shape can become meaningful. To be honest, I think that in the case of M3, having variants is not such a good idea.

I'm pretty sure that nobody has chosen to use the M3 code instead of another one because of its actual shape, but because it encodes the ḫt/wood sign. Changing the shape of the sign will not invalidate any existing encoding, so I can do it without problem.

In other cases, the change might be significant, but I'm pretty sure 90% of occurrences correspond to the version in the Hieroglyphica. For instance, should have the “royal” (straight) beard, not the “divine” beard, as it designates a king. The JSesh shape is simply due to the A40 A40 sign, which has the divine beard, and whose original is a beautiful hieroglyph from Sethy the first's tomb. We reused A40 as the basis for other seated gods and kings signs. The funny thing is that the Hieroglyphica has the same error in reverse, with A40 having the “wrong” beard.

There might be a few cases, for instance in Osirian contexts, where a sign with the exact shape of might perhaps be used (I don't have actual evidence at end right now). But it's a question of compromise. All choices have shortcomings:

• the sign as is has a wrong shape for most occurrences (if not all) occurrences;
• adding a new code with the correct shape would be a worse solution, as it would make all encodings which use A41 wrong;
• changing the current shape to use the correct one might break a few encodings, but it's doubtful. If the current shape does really exist, a new code could be created for it, provided that the change is documented.

The only good approach is the last one. It entails however that we need to document changes in the JSesh encoding.

In other cases, the change will be a breaking change. For instance, JSesh currently has swapped B1J and B1K from the Manuel de Codage. I could decide to keep the current shapes, considering that it's likely that texts have been encoded with them. But the IFAO and an number of scholars use JSesh with fonts derived from the original Winglyph/Hieroglyphica shapes, and have probably the original shapes. As the IFAO is probably a large source of Ptolemaic texts, I consider that the least problematic solution is to change the JSesh shapes to match the original ones, and document the change.

Top priority fixes

Those signs must be fixed quickly. In some case, the problem is a bit minor, but the sign is a frequent one. In other cases, the modification is quite easy to make, and should be done quickly. And finally, some signs are completely wrong.

• A41 : royal beard instead of divine beard;

• A42 : royal beard instead of divine beard;

• A210 : fix the shape of the wood sign;

• A255 : add a small running Apis-bull;

• A357 : both "sw" sign in the same orientation;

• B8G : probably writing implement (ok, that's what I see in the present shape);

• B28 : hands toward the face of the (mourner) figure - note that it's possibly a problematic sign;

• C11L : the arm gesture doesn't fit (basically, the JSesh sign was probably adapted from a sign like C11B C11B who holds rnpt branches). An interesting question is that the features of C11L in the hieroglyphica are :

  • with raised hands;
  • with M4 sign on the head;
  • kneeled with only one leg visible¹;
  • with short wig;
  • naked? having a simple belt?

  If we compare Gardiner's signs with the Hieroglyphica, the Gardiner C11 differs from the Hieroglyphica C11:

  • long wig vs. short wig;
  • long divine? beard;
  • apron vs. possibly only belt;
  • in general, leg posture. Gardiner's C11 back leg is flat on the ground, and the feet sole is turned upward.

  The shape in Gardiner's font corresponds to a detailled sign in Griffith, A Collection of Hieroglyphs, pl. 2, Deir-el-Bahari, chapel of Thothmes I. End wall.

  If we consider the defining features of C11, the fondamental ones are really that it's kneeled, has raised arms, and a M4 sign on the head. it leaves the leg posture, the presence of an apron, and the headress, and the beard as optional features. For four features, it means potentially 16 variants, probably without a specific meaning. This is something I find a bit annoying in the current approaches to sign definition.

  If I compare the Hieroglyphica C11L and C11, I see that the only difference is the absence of the front leg.

  Hence, the new JSesh sign will have the same features as JSesh regular C11, save the leg position.

  (I may consider using Gardiner shape for C11, for the sake of font consistency).

• E8A: fix the sign's hears?

  Actually, the sign's hears are nearly the same as Gardiner's E8*. I need to get a better view on actual witnesses for this sign.

• E8B: we need to fix E8B, which is currently a veal instead of the correct kid. The correct sign is a kid with a ḫtm seal; kind of mix between E31 and E8;

• E9B : the sign should be almost standing

• E154A : momiform shape

• E187 : should have a ankh shaped collar (we create E31B instead);

• F27B : fix tail shape in JSesh.

• G14A: the legs are longer and thiner. The neck is not as close to the wings, and the sign looks balder. Also: G15A, G176A, G177B, G179A , G182A, G183A, G184A, G186D, G187, G283, G284

• G204 : no beard (female)

• G204A : no beard

• G302 : holds a lotus leaf, doesn't feed chicks

• H44 : more curved neck;

• H45 : without the three dots (so, in Hieroglyphica 2000, but apparently not in some versions of Winglyph, according to the online list made by R. Montfort using winglyph).

• I70B : horizontal curves (the current JSesh sign is the same as I70D)

• M3 : fix shape

• M3B : fix shape

• M3E : fix shape

• M8D : four flowers, not three

• N36D : wildly different form; JSesh form is same as N36C

• O19 : move closer to Gardiner's shape?

• P8A : get the same oar shape as hieroglyphica

• P63 : not the same as P66

• Q45 : four feathers

• Q47 : not a bull's head

• S60A : completely different sign? change or keep?

• S65 : damaged sign in JSesh

• T77 : with snake

• V2 : use Gardine's shape (loop above horizontal stroke)

• W17C quite different details

• W18A quite different details

• W59A different vase

• X2 : fix shape a little

• Aa1A add a version of Aa1 with oblique strokes

• Aa23B add transversal lines to vertical poles

Lower priority signs

Those signs should be fixed at some point, but it's not as mandatory in my opinion.

• A13E: make clearer it's an asiatic;
• A194A : holds a short wooden hammer
• E148K : change the knife orientation
• E157A move the ankh in the claws of the sphinx
• P54 : add sundisk to the falcon
• S60 : Uraei on the ram's horns
• T103A : change knife orientation

2. Missing signs

Late versions of the Hieroglyphica have added a number of signs which are not in JSesh. They is no problem in adding them (except that they need to be drawn, that is)

• E148M and E148N
• E207A
• E295
• E296
• E297
• F124B
• G23A
• G31F
• G129D
• G185A
• G266A
• I159A
• I138
• N6E (from Vector Office X)
• O107B is JSesh O308
• O107C is JSesh O308A
• O354
• O355
• R18G
• S204
• S205
• S206
• T100A
• V141
• Y18D
• Z40 (from Vector Office X)
• Z41 (from Vector Office X)
• Aa81

Signs N97A, N97B and N97C are compound signs for N18:X8 with a double stroke y) on the N18, N18:X8 and N102:X8. I'm not sure they deserve a code, but for compatibility reasons it would be nice.

3. Problematic signs

E1

In a first step, I had added E1 in the list of signs to fix. In the Hieroglyphica, the sign has horns similar to those of E2. But Gardiner's sign has has slightly lyra-shaped horns. Renaud de Spens² gives a number of examples from the Old Kingdom to Ramesside times. In all cases listed for actual E1, the horns are lyra-shaped.

This being said, we still face the same problem. It would be very useful to be able to choose if we decide to encode E1 for its linguistic value, or if we pay attention to its actual drawing. Besides, one can find cases where E1 is clearly an ox or a bull, and cases where it's clearly a cow.

My final choice has been to use the lyra shaped horns found in Gardiner. It seems more typical.

A239

This is a typical case of problem we meet when hieroglyphic fonts are extended. When the choice of signs is small, encoders might use the same sign for multiple uses... and in some cases, the ancient scribes could do the same.

Let's take the example of A239 . The original lead fonts of the IFAO has this shape:

Some colleagues complained about my rendering of A239, saying that the sign should be naked, as it is the child-god Ꞽḥy. If we look the Hieroglyphica fonts, and compare them to the IFAO font, we see that:

• there are a number of variants of A239 which are, without doubts, representations of Ꞽḥy (e.g. A240) (A240);
• some of those signs, e.g. A240 should be in the “C” family; Dimitry Meeks' classification (A = men, B=kings and human-shaped deities, C=women, D=non human headed deities) would improve things a little, except perhaps for A239 itself
• the IFAO font has two very similar signs: 30,7 and 30,8. The latter is probably naked, and has somehow more childish proportions, and thus might be Ꞽḥy:

Basically, we have:

• a sign, known from the Old Kingdom onwards, which shows a musician, and which is used as determinative for actual human musicians; the shape in the hieroglyphica corresponds to this sign;
• a sign showing the god Ꞽḥy, without crown. The sign should wear the lock of childhood, and is prototypically naked, at least in the Ptolemaic period. However, we can find clothed variants of the sign:

If (and probably when) we fix the problem, we will need to add a new sign, explicitly representing Ꞽḥy,probably in family A, as a variant of A239 or, more correctly A240. Family “C” would be better, but it would place different versions of Ꞽḥy in different families, which is somehow anoying.

Now, this will break lots of already-encoded texts. Actually, as I suspect one find more occurrences of Ꞽḥy than of “sistrum-playing musician”, it's quite likely that most texts encoded with the current A239 code will in fact refer to this “new” variant of A240. I'm afraid I have no good solution for this. The best I can probably do is to create a section in JSesh web site dedicated to possibly breaking changes in the encoding.

Once again, the breaking change here will be the result of the mere addition of a sign, without any change to existing signs.

Wrong shape, but existing sign?

In some cases, JSesh has wrong shapes for a sign... but instances of this wrong shape could exist.

Take for instance A302A. Its current JSesh shape is . But it should in fact wear the red crown.

Now, I don't know if someone, somewhere, has found an occurrence of the shape used in JSesh. If this is the case, fixing the encoding problem might cause unwanted changes in existing encoded texts.

A6J

This might be also the case for A6J (but it's quite theoretical)

• A6J (completely different posture);

A92

The case of A92 is more serious:

• A92 : attitude of proskynesis, not a fallen ennemy;

“Fallen ennemies” are one of those categories of signs where monumental texts go wild in variation. And it's quite likely that someone has use A92 to actually encode a fallen ennemy. Actually, this is the case. In JSesh texts, and in the Ramses database, I find A92 used for :

• determinative of sn-tꜣ (definitly proskynesis);
• determinative of wr:r wr, Qadesh, L1, §11 (somehow ambiguous);
• determinative of h-A-W-A92, Qadesh L1, §142 (definitly fallen ennemy?).

A closer look at the uses of the sign and its actual shapes is really needed here before we take any decision.

C105 and Variants

C105 and variants, for instance, are somehow problematic, and for a fundamental reason in hieroglyphic fonts. Basically, very small details are difficult to see. And without an explicit description of the sign, the choice might be hard to make. A very close look at the signs hints toward two family of variants for a "wounded Sethian-looking ennemy". Some of them have the head of an ass, other have the head of the Seth animal. It's quite difficult to distinguish the two at the sign of a font glyph, so it's left for a written description.

D251

D251 should be holding the ḫrp scepter, not the ḏsr one. According to Cauville, Fond Hiéroglyphique, the sign has the value ḏzr (DIII, 39,8). Hence, the object is indeed the nḥb.t wand, seen in the Dsr sign.

It should perhaps be made a little less round.

M3A

M3A is a variant of the branch in the Hieroglyphica (which M. Thomas had drawn for the JSesh sign library as ) (see here).

The choice of shape for M3A in JSesh font comes from variants in Bob Richmond's Inscribe (which predates the Hieroglyphica). Gardiner, in the Supplement to the Catalogue of the Egyptian Hieroglyphic Printing Type, Showing Acquisitions to December 1953, p. 4, uses the code M3* for , and the original rationale in the Manuel de Codage was that suffixes A,B,C... were to be used for *, **, ***... found in Gardiner list for variants. Thus, M3A is the Manuel de Codage equivalent of M3*. At some point, I decided to give a priority to the codes in Gardiner's font. Now, this is reasonable for the list in the grammar itself. Regarding the later extensions, I'm not sure I have taken the right decision.

Anyway, this is quite a problem, as the JSesh shape corresponds to a relatively usual spelling of m-ḫt. The problem is that fixing the list to be true to the Hieroglyphica would be likely to create errors in existing texts. I have no really good solution here.

Blunders in JSesh fonts

There are also a number of blunder in the JSesh fonts which can't really be fixed without breaking texts.

For instance :

• B1J and B1K : I inverted those two signs viz. the Hieroglyphica. However, it has been done quite a few years ago, and texts have been encoded with it. On the other hand, the fonts used by the IFAO with JSesh have the correct shapes.

4. Conclusion

• The JSesh web site should contain a section about encoding changes. This section should be prominently featured on the web site;
• code-breaking changes should be indicated as such;
• in a few cases, we might choose to keep the JSesh version - but document it.