This table summarizes UTF-8’s variable-width encoding scheme.

After noticing that not all the available 2-byte encodings are used, I extended the table as shown below, and saw that this was the case for 3-byte and 4-byte encodings as well (delta > 0).

| number of bytes | bits available for encodings | number of available encodings | number of codepoints                                | delta  = # avail encodings - # codepoints |
|-----------------|------------------------------|-------------------------------|-----------------------------------------------------|-------------------------------------------|
| 1               |          7                   |  2^7 =       128              |       0x80 -    0x00 =             8*16 =       128 |       128 -       128 =         0         |
| 2               | 5+1*6 = 11                   | 2^11 =     2,048              |     0x800 -    0x080 =    8*16^2 - 8*16 =     1,920 |     2,048 -     1,920 =       128         |
| 3               | 4+2*6 = 16                   | 2^16 =    65,536              |  0x1_0000 -    0x800 =    16^4 - 8*16^2 =    63,488 |    65,336 -    63,488 =     2,048         |
| 4               | 3+3*6 = 21                   | 2^21 = 2,097,152              | 0x11_0000 - 0x1_0000 = 0x10_0000 = 16^5 = 1,048,576 | 2,097,152 - 1,048,576 = 1,048,576         |

Is there a reason that so many encodings are unused?