source-han-serif icon indicating copy to clipboard operation
source-han-serif copied to clipboard

Adjustment suggestion to HK/TW's ⿰⺼ component

Open tamcy opened this issue 2 years ago • 6 comments

Preface

This issue is about the component component(⺼), specifically about the placement of the two strokes in the middle, i.e. component-middle(冫) in Source Han Serif v2.000.

The component is a variant form of the radical 肉.

For CN/JP/KR, ⺼ is amost always written as 月 when placed on the left/right or bottom of the glyph. For TW and HK, the two horizontal strokes become "冫" when it is on the left. When placed at the right or bottom, TW still uses the "冫" form, while HK glyphs are written like 月.

region-diff

This issue is focused on the "⺼(⿴⺆冫)" form adopted by TW and HK, and affects only with the composition ⿰⺼, like 肥, 俷, 蜰 or 葋. ⺼ on the right (like 胡) and at the bottom (like 胃, 骨, 肖) are not affected.

The problem

For most (if not all) of the characters with the composition ⿰⺼, the placement of the two middle strokes 冫 seems always too high.

When I tried to use Source Han Serif to typeset a document, the placement of 冫 somewhat surprised me - I immediately thought "wait, isn't it a bit too high"?

I then check the glyphs having ⺼ on the left and found this is not an isolated case. Here are some examples that the middle strokes are obviously too "high" to my eyes. You may also look at the character 肋 above .

high_examples

I did an extremely rough calculation. For the horizontal version 月, the height of the three whitespace is about 1 : 1 : 1.5. For the TW/HK version ⺼, I measured the heights of the left and right side of the whitespace area and took the average, and the ratio is about 1 : 1 : 1.95. Obviously there is too much whitespace at the bottom. I'm not good at maths so just take it as a grain of salt. But practically, the glyph looks more "natural" to me when I shifted the two middle strokes 冫 downwards (and it seems a little more space could be assigned to the middle whitespace).

adjusted_form

So I'd like to suggest revisiting the characters with with the structure ⿰⺼, and make adjustments to the placement to the strokes of 冫 where neccessary. I am illustrating the issue using only the ExtraLight master, but I suggest to also review the Heavy master glyphs.

Other notes

As said, this issue doesn't apply to characters with ⺼ on the right. An example is 湖 (U+6E56). While I think the horizontal placement of the middle component 冫 (esp. the dot) could be further optimized, the vertical placement of 冫 looks fine to me:

ok-compare

The reason is easy to tell when comparing the vertical placement of the 冫 component of the TW glyph (black) against the two horizontal strokes of the JP glyph (green):

compare

This may serve as a modification reference if you decide to tackle this issue (just the vertical spacing, not the horizontal placement).

Affected codepoints

Click to see the list of affected codepoints and glyphs:
# Codepoint Character TW HK
1 U+3B39 uni3B39-HK
2 U+43D9 uni43D9-HK
3 U+43DF uni43DF-HK
4 U+43F0 uni43F0-HK
5 U+43F2 uni43F2-HK
6 U+4401 uni4401-HK
7 U+4402 uni4402-HK
8 U+4413 uni4413-HK
9 U+4425 uni4425-HK
10 U+442D uni442D-HK
11 U+4FF7 uni4FF7-TW
12 U+6721 uni6721-TW uni6721-HK
13 U+6DDD uni6DDD-TW
14 U+7B4B uni7B4B-TW
15 U+808A uni808A-TW
16 U+808B uni808B-TW
17 U+808C uni808C-TW
18 U+8090 uni8090-TW
19 U+8092 uni8092-TW
20 U+8095 uni8095-TW
21 U+8098 uni8098-TW
22 U+809A uni809A-TW
23 U+809B uni809B-TW
24 U+809C uni809C-TW
25 U+809D uni809D-TW
26 U+809F uni809F-HK
27 U+80A1 uni80A1-TW
28 U+80A2 uni80A2-TW
29 U+80A3 uni80A3-TW
30 U+80A5 uni80A5-TW
31 U+80A7 uni80A7-HK
32 U+80AA uni80AA-TW
33 U+80AB uni80AB-TW
34 U+80AD uni80AD-TW
35 U+80AE uni80AE-TW
36 U+80B1 uni80B1-TW
37 U+80B5 uni80B5-TW
38 U+80B6 uni80B6-HK
39 U+80B7 uni80B7-HK
40 U+80B8 uni80B8-TW uni80B8-HK
41 U+80BA uni80BA-TW
42 U+80C2 uni80C2-TW
43 U+80C5 uni80C5-TW
44 U+80C6 uni80C6-HK
45 U+80C7 uni80C7-TW
46 U+80C8 uni80C8-TW
47 U+80C9 uni80C9-TW
48 U+80CA uni80CA-CN*
49 U+80CD uni80CD-TW
50 U+80CE uni80CE-TW
51 U+80CF uni80CF-TW
52 U+80D0 uni80D0-CN*
53 U+80D1 uni80D1-TW
54 U+80D5 uni80D5-TW
55 U+80D6 uni80D6-TW
56 U+80D7 uni80D7-TW
57 U+80D8 uni80D8-TW
58 U+80D9 uni80D9-TW
59 U+80DA uni80DA-TW uni80DA-HK
60 U+80DB uni80DB-TW
61 U+80DC uni80DC-TW
62 U+80DD uni80DD-TW
63 U+80DE uni80DE-TW
64 U+80E0 uni80E0-TW
65 U+80E3 uni80E3-TW
66 U+80E6 uni80E6-TW
67 U+80ED uni80ED-TW
68 U+80EF uni80EF-TW
69 U+80F0 uni80F0-TW
70 U+80F1 uni80F1-TW
71 U+80F2 uni80F2-TW
72 U+80F3 uni80F3-TW
73 U+80F4 uni80F4-TW
74 U+80F5 uni80F5-TW
75 U+80F6 uni80F6-HK
76 U+80F8 uni80F8-TW
77 U+80F9 uni80F9-TW
78 U+80FA uni80FA-TW
79 U+80FB uni80FB-TW
80 U+80FC uni80FC-TW
81 U+8101 uni8101-TW
82 U+8102 uni8102-TW
83 U+8103 uni8103-HK
84 U+8106 uni8106-TW
85 U+8107 uni8107-HK
86 U+8108 uni8108-TW
87 U+8109 uni8109-HK
88 U+8115 uni8115-TW
89 U+8116 uni8116-TW
90 U+8117 uni8117-HK
91 U+8118 uni8118-TW
92 U+8119 uni8119-TW
93 U+811A uni811A-HK
94 U+811B uni811B-TW
95 U+811D uni811D-TW
96 U+811E uni811E-TW
97 U+811F uni811F-TW
98 U+8121 uni8121-TW
99 U+8122 uni8122-TW
100 U+8124 uni8124-TW
101 U+8125 uni8125-TW
102 U+8127 uni8127-TW uni8127-HK
103 U+812A uni812A-HK
104 U+812B uni812B-TW uni812B-HK
105 U+812C uni812C-TW
106 U+812D uni812D-TW
107 U+812F uni812F-TW
108 U+8130 uni8130-TW
109 U+8131 uni8131-HK
110 U+8134 uni8134-HK
111 U+8139 uni8139-TW
112 U+813A uni813A-TW
113 U+813D uni813D-TW
114 U+813E uni813E-TW
115 U+8142 uni8142-HK
116 U+8143 uni8143-TW
117 U+8144 uni8144-TW
118 U+8146 uni8146-TW
119 U+8147 uni8147-TW
120 U+8148 uni8148-HK
121 U+814A uni814A-TW
122 U+814B uni814B-TW
123 U+814C uni814C-TW uni814C-HK
124 U+814D uni814D-TW
125 U+814F uni814F-TW
126 U+8151 uni8151-TW
127 U+8152 uni8152-TW
128 U+8153 uni8153-TW
129 U+8154 uni8154-TW
130 U+8155 uni8155-TW
131 U+815B uni815B-TW
132 U+815C uni815C-TW uni815C-HK
133 U+815E uni815E-TW
134 U+8160 uni8160-TW
135 U+8161 uni8161-TW
136 U+8162 uni8162-TW
137 U+8164 uni8164-TW
138 U+8165 uni8165-TW
139 U+8166 uni8166-TW
140 U+8167 uni8167-TW uni8167-HK
141 U+8169 uni8169-TW
142 U+816B uni816B-TW
143 U+816C uni816C-HK
144 U+816D uni816D-HK
145 U+816E uni816E-TW
146 U+816F uni816F-TW
147 U+8170 uni8170-TW
148 U+8171 uni8171-TW
149 U+8172 uni8172-TW
150 U+8173 uni8173-TW
151 U+8174 uni8174-TW
152 U+8176 uni8176-TW
153 U+8177 uni8177-TW
154 U+8178 uni8178-TW
155 U+8179 uni8179-TW uni8179-HK
156 U+817A uni817A-TW
157 U+817D uni817D-HK
158 U+817F uni817F-TW
159 U+8180 uni8180-TW
160 U+8183 uni8183-TW
161 U+8184 uni8184-HK
162 U+8186 uni8186-TW
163 U+8187 uni8187-TW
164 U+8188 uni8188-TW
165 U+8189 uni8189-TW
166 U+818A uni818A-TW
167 U+818C uni818C-TW uni818C-HK
168 U+818D uni818D-TW
169 U+8193 uni8193-HK
170 U+8195 uni8195-TW
171 U+8197 uni8197-TW
172 U+8198 uni8198-TW uni8198-HK
173 U+8199 uni8199-TW
174 U+819B uni819B-TW
175 U+819C uni819C-TW
176 U+819D uni819D-TW
177 U+819E uni819E-TW
178 U+819F uni819F-TW
179 U+81A0 uni81A0-TW
180 U+81A2 uni81A2-TW uni81A2-HK
181 U+81A3 uni81A3-TW
182 U+81A6 uni81A6-TW uni81A6-HK
183 U+81A7 uni81A7-CN*
184 U+81A8 uni81A8-TW
185 U+81A9 uni81A9-TW
186 U+81AA uni81AA-HK
187 U+81AB uni81AB-TW
188 U+81AC uni81AC-TW
189 U+81AE uni81AE-TW
190 U+81B0 uni81B0-TW
191 U+81B1 uni81B1-TW
192 U+81B2 uni81B2-TW
193 U+81B3 uni81B3-TW
194 U+81B4 uni81B4-TW
195 U+81B5 uni81B5-TW
196 U+81B7 uni81B7-TW
197 U+81B9 uni81B9-TW
198 U+81BB uni81BB-TW
199 U+81BC uni81BC-TW
200 U+81BD uni81BD-TW
201 U+81BE uni81BE-TW uni81BE-HK
202 U+81BF uni81BF-TW
203 U+81C1 uni81C1-HK
204 U+81C3 uni81C3-TW
205 U+81C4 uni81C4-TW
206 U+81C5 uni81C5-TW
207 U+81C6 uni81C6-TW
208 U+81C7 uni81C7-TW
209 U+81C8 uni81C8-HK
210 U+81C9 uni81C9-TW
211 U+81CA uni81CA-TW uni81CA-HK
212 U+81CC uni81CC-TW
213 U+81CD uni81CD-TW uni81CD-HK
214 U+81CF uni81CF-TW
215 U+81D0 uni81D0-TW
216 U+81D1 uni81D1-TW
217 U+81D2 uni81D2-TW
218 U+81D5 uni81D5-TW
219 U+81D7 uni81D7-TW
220 U+81D8 uni81D8-TW uni81D8-HK
221 U+81D9 uni81D9-TW
222 U+81DA uni81DA-TW
223 U+81DB uni81DB-TW
224 U+81DE uni81DE-TW
225 U+81DF uni81DF-TW
226 U+81E2 uni81E2-TW
227 U+8409 uni8409-TW
228 U+8444 uni8444-TW
229 U+844B uni844B-TW
230 U+8730 uni8730-TW
231 U+9721 uni9721-HK
232 U+9722 uni9722-TW
233 U+2112F 𡄯 u2112F-HK

I understand this is not a trival modification. I also tried to see if it is possible to reduce its size by only including frequently used characters. This reduces the number of items to ~140, and the number of glyphs to ~150:

Click to see the reduced list, which excludes infrequently used characters:
# Codepoint Character TW HK
1 U+3B39 uni3B39-HK
3 U+43DF uni43DF-HK
6 U+4401 uni4401-HK
13 U+6DDD uni6DDD-TW
14 U+7B4B uni7B4B-TW
16 U+808B uni808B-TW
17 U+808C uni808C-TW
20 U+8095 uni8095-TW
21 U+8098 uni8098-TW
22 U+809A uni809A-TW
23 U+809B uni809B-TW
25 U+809D uni809D-TW
26 U+809F uni809F-HK
27 U+80A1 uni80A1-TW
28 U+80A2 uni80A2-TW
30 U+80A5 uni80A5-TW
32 U+80AA uni80AA-TW
33 U+80AB uni80AB-TW
34 U+80AD uni80AD-TW
35 U+80AE uni80AE-TW
36 U+80B1 uni80B1-TW
38 U+80B6 uni80B6-HK
41 U+80BA uni80BA-TW
42 U+80C2 uni80C2-TW
44 U+80C6 uni80C6-HK
49 U+80CD uni80CD-TW
50 U+80CE uni80CE-TW
55 U+80D6 uni80D6-TW
56 U+80D7 uni80D7-TW
59 U+80DA uni80DA-TW uni80DA-HK
60 U+80DB uni80DB-TW
61 U+80DC uni80DC-TW
62 U+80DD uni80DD-TW
63 U+80DE uni80DE-TW
67 U+80ED uni80ED-TW
68 U+80EF uni80EF-TW
69 U+80F0 uni80F0-TW
70 U+80F1 uni80F1-TW
72 U+80F3 uni80F3-TW
73 U+80F4 uni80F4-TW
74 U+80F5 uni80F5-TW
75 U+80F6 uni80F6-HK
76 U+80F8 uni80F8-TW
78 U+80FA uni80FA-TW
80 U+80FC uni80FC-TW
82 U+8102 uni8102-TW
84 U+8106 uni8106-TW
85 U+8107 uni8107-HK
86 U+8108 uni8108-TW
87 U+8109 uni8109-HK
89 U+8116 uni8116-TW
90 U+8117 uni8117-HK
91 U+8118 uni8118-TW
93 U+811A uni811A-HK
94 U+811B uni811B-TW
96 U+811E uni811E-TW
99 U+8122 uni8122-TW
102 U+8127 uni8127-TW uni8127-HK
104 U+812B uni812B-TW uni812B-HK
105 U+812C uni812C-TW
107 U+812F uni812F-TW
109 U+8131 uni8131-HK
111 U+8139 uni8139-TW
113 U+813D uni813D-TW
114 U+813E uni813E-TW
116 U+8143 uni8143-TW
120 U+8148 uni8148-HK
121 U+814A uni814A-TW
122 U+814B uni814B-TW
123 U+814C uni814C-TW uni814C-HK
124 U+814D uni814D-TW
126 U+8151 uni8151-TW
128 U+8153 uni8153-TW
129 U+8154 uni8154-TW
130 U+8155 uni8155-TW
138 U+8165 uni8165-TW
139 U+8166 uni8166-TW
141 U+8169 uni8169-TW
142 U+816B uni816B-TW
144 U+816D uni816D-HK
145 U+816E uni816E-TW
146 U+816F uni816F-TW
147 U+8170 uni8170-TW
148 U+8171 uni8171-TW
150 U+8173 uni8173-TW
151 U+8174 uni8174-TW
152 U+8176 uni8176-TW
153 U+8177 uni8177-TW
154 U+8178 uni8178-TW
155 U+8179 uni8179-TW uni8179-HK
156 U+817A uni817A-TW
157 U+817D uni817D-HK
158 U+817F uni817F-TW
159 U+8180 uni8180-TW
160 U+8183 uni8183-TW
164 U+8188 uni8188-TW
165 U+8189 uni8189-TW
166 U+818A uni818A-TW
167 U+818C uni818C-TW uni818C-HK
168 U+818D uni818D-TW
170 U+8195 uni8195-TW
172 U+8198 uni8198-TW uni8198-HK
174 U+819B uni819B-TW
175 U+819C uni819C-TW
176 U+819D uni819D-TW
177 U+819E uni819E-TW
178 U+819F uni819F-TW
179 U+81A0 uni81A0-TW
181 U+81A3 uni81A3-TW
182 U+81A6 uni81A6-TW uni81A6-HK
184 U+81A8 uni81A8-TW
185 U+81A9 uni81A9-TW
193 U+81B3 uni81B3-TW
195 U+81B5 uni81B5-TW
198 U+81BB uni81BB-TW
199 U+81BC uni81BC-TW
200 U+81BD uni81BD-TW
201 U+81BE uni81BE-TW uni81BE-HK
202 U+81BF uni81BF-TW
203 U+81C1 uni81C1-HK
204 U+81C3 uni81C3-TW
205 U+81C4 uni81C4-TW
207 U+81C6 uni81C6-TW
210 U+81C9 uni81C9-TW
211 U+81CA uni81CA-TW uni81CA-HK
212 U+81CC uni81CC-TW
213 U+81CD uni81CD-TW uni81CD-HK
214 U+81CF uni81CF-TW
216 U+81D1 uni81D1-TW
219 U+81D7 uni81D7-TW
220 U+81D8 uni81D8-TW uni81D8-HK
225 U+81DF uni81DF-TW
226 U+81E2 uni81E2-TW
233 U+2112F 𡄯 u2112F-HK
  • For U+80CA 胊, U+80D0 胐 and U+81A7 膧, the CN glyph adopts the "⺼/⿴⺆冫" form instead of 月. The design is a little bit different from TW/HK's ⺼, and

This is not surprising, because many the characters like 腦=brain, 腳=leg, 臉=face, 臟腑=organs, 膨脹=expand, 脂肪/肥胖=fat, 脫=take off (clothes etc.), escape, so using these characters in an article is unavoidable.

tamcy avatar Nov 14 '21 03:11 tamcy

Like https://github.com/adobe-fonts/source-han-serif/issues/127 this one has also been bothering me.

punchcutter avatar Nov 15 '21 06:11 punchcutter

It should be noted that ⺼ in left and bottom should touch the left and right stroke according to The Dictionary of Chinese Variant Form (異體字字典) and Source Han Serif did not met the requirement for bottom ⺼. For ⺼ on the right the MOE Sung demonstrate that most are touching, and Source Han Serif appears to be touching too. image

Affected characters as viewed in Regular: (201 characters) ⺼肓肖肙肩肯育肴胃胄背胡胤胥能脅脊脣脩腎膂膏膺臀臂佾俏偝削厭厴哨唷啃喟嗋堉塉壓娟婿媦嬮宵尳屑峭崤帩弰悁悄愲愶態捎捐掮掯揟揱揹搚搰散旓梋梢楈榍榾樇殽消涓淆淯渭湑湚滑滫漘潲澈焆煟熁熊犤狷猏猒猾琄痟瘠睄睊矲硝磭稍稰筋筲箾糈絹綃縃縎罥罷羆艄莦菺萷葫蓇蓨蕱藣蘛蛸蜎蜰蝑蝟螖螚裐褙誚諝謂趙踃蹐輎逍郩鄁酳醑銷鋗錥錹霄霰靨鞘鞙顅顝餚饜駽骨骫骭骯骰骱骳骴骷骸骹骻骼骾骿髀髁髂髆髇髊髍髏髐髑髒髓體髕髖髾鬍魈魘鮹鵑鵳鶦鶺鶻黶 image

NightFurySL2001 avatar Nov 20 '21 09:11 NightFurySL2001

@NightFurySL2001 I am not sure why you are still keeping your eyes on this "to touch or not to touch" issue after reading the official reply you got yesterday. And the concern you posted here and in other issue(s) may have been well answered here, and here.

tamcy avatar Nov 20 '21 10:11 tamcy

Characters containing the ⺼component overwhelmingly apply to glyphs unique to TW or glyphs shared only between TW and HK, so there is little effect on the glyph count if such "noncompliance" was rectified.

However, as compliance will yield an objectively horrible result, I believe proposing so is a form of malicious compliance. 😆

hfhchan avatar Nov 20 '21 10:11 hfhchan

I would like to point out a few things:

  1. the current ⺼ only applies for TW and HK, so no new glyphs will be required for this operation;

  2. this edit is trivial by just stretching the 冫 across and touching;

  3. both left and (most) right ⺼ component already has a touching form and only bottom ⺼does not touch the left/right stroke, causing inconsistency in the font itself. image

  4. the comment provided to SHSans does not fit this issue as 1. the example is not mentioned in any documents in Taiwan MOE; 2. these glyphs are non-sharable already due to the design of ⺼ itself. Adobe however should consult with the designers whether it is ok to fix this issue.

NightFurySL2001 avatar Nov 20 '21 11:11 NightFurySL2001

While I don't think it is a good idea to force the middle strokes to touch the outer component, I agree to leave it to Adobe's designer to decide. And just FYI, Taiwan MoE's official Sung (Serif) and Kai (Script) fonts also exhibit such "inconsistencies": meat

tamcy avatar Nov 20 '21 13:11 tamcy