Since JDK 1.2 the windows font properties file has specified to use a Japanese font for the base monospaced font
in order that there is an exact 2:1 width ratio for the 1/2 width latin chars relative to full width Kanji
See bug https://bugs.openjdk.org/browse/JDK-4159896

In it's current form it looks like this - spelling out the sequence for each logical font so monospaced is as required
===
sequence.serif.windows-31j=alphabetic,japanese,dingbats,symbol
sequence.sansserif.windows-31j=alphabetic,japanese,dingbats,symbol
sequence.monospaced.windows-31j=japanese,alphabetic,dingbats,symbol
sequence.dialog.windows-31j=alphabetic,japanese,dingbats,symbol
sequence.dialoginput.windows-31j=alphabetic,japanese,dingbats,symbol

====

[NB There was similar work even earlier - JDK1.1.7 - on Solaris : https://bugs.openjdk.org/browse/JDK-4101757

In JDK 7 there was a request to add support UTF-8.ja and it was added under https://bugs.openjdk.org/browse/JDK-6856390
and the diff was just this and it is still the same today
===
diff --git a/jdk/src/windows/classes/sun/awt/windows/fontconfig.properties b/jdk/src/windows/classes/sun/awt/windows/fontconfig.properties
index a82b82422b..b3aef483fd 100644
--- a/jdk/src/windows/classes/sun/awt/windows/fontconfig.properties
+++ b/jdk/src/windows/classes/sun/awt/windows/fontconfig.properties
@@ -211,6 +211,7 @@ sequence.dialog.x-MS950-HKSCS=alphabetic,chinese-ms950,chinese-hkscs,dingbats,sy
 sequence.dialoginput.x-MS950-HKSCS=alphabetic,chinese-ms950,chinese-hkscs,dingbats,symbol,chinese-ms950-extb
 
 sequence.allfonts.UTF-8.hi=alphabetic/1252,devanagari,dingbats,symbol
+sequence.allfonts.UTF-8.ja=alphabetic,japanese,devanagari,dingbats,symbol
====

So clearly in this case we won't get the right width ratio and it looks as if we need to enumerate all the cases again.