Unicode defines a huge set of characters for use with computer displays and printers. Some characters denote entire words (as in many Asian languages), alphabetic letters (as in many European languages), digits, and punctuation; while other characters have sundry purposes. This report examines the Unicode characters in two categories: box drawing and block elements, which here are for convenience aggregated as geometric.

The hardware and software of most modern general-purpose computers support control of individual display pixels in a rectangular array known as a raster, and this capability is practically required for a graphical user interface. However, embedded computers often entail relatively simple output, and might perform satisfactorily with a simpler display constructed as a two-dimensional array of perhaps a thousand characters rather than a million pixels. Also, many computers built before 1990 used character-based displays because of the higher costs of hardware at that time; a standard size was 25 lines of 80 characters each, using a monospaced font.

One symbol can be displayed within each character cell, which is typically implemented as a small array of pixels. On some machines, the height of a pixel exceeds its width; this design reduces the number of pixels while preserving the taller-than-wide aspect exhibited by most letters and digits in traditional typography. However, there is a tendency in modern equipment to make pixels square.

A practical minimum for text in the English language is a character size of eight pixels wide by eight pixels high; some other languages need more pixels for legibility. In the following example, which uses tall pixels, each character resides in such an eight-by-eight region, but the rightmost column and bottommost row of pixels are left unused to provide spacing between characters. Hence the effective size is seven by seven. For this illustration, pixels not used by the characters are rendered in a checkerboard pattern to show the extent of each.

If only Arabic numerals are to be shown, a width of four pixels and a height of six pixels (with a usable size of three by five) will suffice. In the following example, square pixels happen to be used.

Character-based displays can be very efficient because the geometric details of how to draw each character are stored in the display device itself, and are compact: the pattern for each character would require no more than eight bytes of storage for the alphabetic example above, and three bytes for the numeric. To make a character appear, the controlling computer need send merely a brief code, often only eight bits, and then the display figures out how to draw the character. A tradeoff is that the user often has limited choice of factors that affect the appearance of the font, such as face, size or color.

More generally, nothing in Unicode requires that the display be organized into pixels; some other technology could instead be used as long as the characters have a suitable appearance. For instance vector-based displays have been built, but are rare today; for hard copy, plotters are more frequently seen.

The earliest devices generally had light-colored (white and green were common) characters on a dark background. The IBM 3279 display, frequently seen in the 1980s, was among the first to give a choice of colors, producing red, green, blue, yellow, cyan, magenta and white figures on a black background. Each character could be a different color, but colors were never mixed within a character.

Many displays have offered:

• inverse video, with dark characters on a light background
• underscoring
• blinking characters, at a rate of roughly one cycle per second.

Meanwhile, printers will almost always yield dark characters on a paper-white background.

By the early 1980's, many desktop computers were extending their character sets by adding a variety of geometric characters. The most basic were straight lines, right angles, tees and crosses, but many other shapes were included. These figures could be arranged on the display to divide the screen into sections or to draw simple pictures. Early examples are PETSCII, code page 437. Also Videotex, an early communications service, defined some geometric characters.

Motivating the establishment of Unicode was a lack of standardization among manufacturers' character sets, not only in the geometric characters but also in Latin alphabet letters with diacritical marks (such as accents and umlauts), and in non-Latin alphabets (such as Greek and Cyrillic). Moreover in some languages (such as Chinese) words cannot be decomposed into sequences of letters, but rather are an integral symbols; Unicode addresses these also.

Here is an demonstration of why a monospaced font is essential when using these geometic characters. This first screenshot has twelve characters per line, and is rendered in the font Deja Vu Sans Mono, wherein all characters are the same width:

In the well-known Times New Roman font each character is of its natural width, which on the one hand makes for easy reading of large blocks of prose, but which on the other hand makes for difficulty in aligning columns of text. This font is used in the next screenshot, where alignment of the box drawing characters fails severely even though precisely the same characters are used as the previous screenshot:

Adding or deleting characters gives at best an approximate fix:

A display that is arranged as a grid of characters automatically gives monospacing and averts this problem.

Unicode's box drawing characters and block elements, in table one below, resemble many of the early geometric characters. Exhibited are sample glyphs for the 160 Unicode characters numbered 2500-259F (in hexadecimal) or 9472-9631 (in decimal). Although the standards (box drawing, block elements) do indicate the general appearance of the glyphs, implementors are granted flexibility in rendering them.

For each character of table one, four items appear:

• Its glyph, as gif file.
• Its Unicode number in hexadecimal (2xxx)
• Its Unicode number in decimal (9xxx)
• Its rendering according to the font employed by the user's browser.

The gif images are necessary for two reasons. First, many fonts support Unicode poorly. Second, some alternative glyphs that are not found in Unicode will appear later in the discussion, and the gifs allow all to be drawn in the same format.

In the charts of the Unicode standard the glyphs are shown as black on a white background, but this report uses instead a gray background to show how the figure meets the cell boundary. In the actual display, each pixel is either fully on or fully off, because Unicode does not support intermediate levels of brightness. In order to make details clearer, the gifs below (mostly 24 × 48 pixels) are considerably larger than what is usually found on computer displays.

Table one.
Box drawing	2500 9472 ─	2501 9473 ━	2502 9474 │	2503 9475 ┃	2504 9476 ┄	2505 9477 ┅	2506 9478 ┆	2507 9479 ┇	2508 9480 ┈	2509 9481 ┉	250A 9482 ┊	250B 9483 ┋	250C 9484 ┌	250D 9485 ┍	250E 9486 ┎	250F 9487 ┏
	2510 9488 ┐	2511 9489 ┑	2512 9490 ┒	2513 9491 ┓	2514 9492 └	2515 9493 ┕	2516 9494 ┖	2517 9495 ┗	2518 9496 ┘	2519 9497 ┙	251A 9498 ┚	251B 9499 ┛	251C 9500 ├	251D 9501 ┝	251E 9502 ┞	251F 9503 ┟
	2520 9504 ┠	2521 9505 ┡	2522 9506 ┢	2523 9507 ┣	2524 9508 ┤	2525 9509 ┥	2526 9510 ┦	2527 9511 ┧	2528 9512 ┨	2529 9513 ┩	252A 9514 ┪	252B 9515 ┫	252C 9516 ┬	252D 9517 ┭	252E 9518 ┮	252F 9519 ┯
	2530 9520 ┰	2531 9521 ┱	2532 9522 ┲	2533 9523 ┳	2534 9524 ┴	2535 9525 ┵	2536 9526 ┶	2537 9527 ┷	2538 9528 ┸	2539 9529 ┹	253A 9530 ┺	253B 9531 ┻	253C 9532 ┼	253D 9533 ┽	253E 9534 ┾	253F 9535 ┿
	2540 9536 ╀	2541 9537 ╁	2542 9538 ╂	2543 9539 ╃	2544 9540 ╄	2545 9541 ╅	2546 9542 ╆	2547 9543 ╇	2548 9544 ╈	2549 9545 ╉	254A 9546 ╊	254B 9547 ╋	254C 9548 ╌	254D 9549 ╍	254E 9550 ╎	254F 9551 ╏
	2550 9552 ═	2551 9553 ║	2552 9554 ╒	2553 9555 ╓	2554 9556 ╔	2555 9557 ╕	2556 9558 ╖	2557 9559 ╗	2558 9560 ╘	2559 9561 ╙	255A 9562 ╚	255B 9563 ╛	255C 9564 ╜	255D 9565 ╝	255E 9566 ╞	255F 9567 ╟
	2560 9568 ╠	2561 9569 ╡	2562 9570 ╢	2563 9571 ╣	2564 9572 ╤	2565 9573 ╥	2566 9574 ╦	2567 9575 ╧	2568 9576 ╨	2569 9577 ╩	256A 9578 ╪	256B 9579 ╫	256C 9580 ╬	256D 9581 ╭	256E 9582 ╮	256F 9583 ╯
	2570 9584 ╰	2571 9585 ╱	2572 9586 ╲	2573 9587 ╳	2574 9588 ╴	2575 9589 ╵	2576 9590 ╶	2577 9591 ╷	2578 9592 ╸	2579 9593 ╹	257A 9594 ╺	257B 9595 ╻	257C 9596 ╼	257D 9597 ╽	257E 9598 ╾	257F 9599 ╿
Block elements	2580 9600 ▀	2581 9601 ▁	2582 9602 ▂	2583 9603 ▃	2584 9604 ▄	2585 9605 ▅	2586 9606 ▆	2587 9607 ▇	2588 9608 █	2589 9609 ▉	258A 9610 ▊	258B 9611 ▋	258C 9612 ▌	258D 9613 ▍	258E 9614 ▎	258F 9615 ▏
	2590 9616 ▐	2591 9617 ░	2592 9618 ▒	2593 9619 ▓	2594 9620 ▔	2595 9621 ▕	2596 9622 ▖	2597 9623 ▗	2598 9624 ▘	2599 9625 ▙	259A 9626 ▚	259B 9627 ▛	259C 9628 ▜	259D 9629 ▝	259E 9630 ▞	259F 9631 ▟

The display device should insert no gap between characters, or breaks will appear in what ought to be continuous lines.

Wrong	Right

Unicode does not attempt to supply all geometric characters that might be needed for all purposes, rather providing a moderately sized set that fulfills many needs, section 15.7 of the standard giving a rationale. Applications requiring fine control of appearance will need to use displays that allow control of individual pixels.

As an example of the variation in acceptable glyph appearance, the next table shows ten ways in which character 256C is rendered in the bitmapped fonts of Microsoft Windows XP SP2. Note that some are asymmetric.

Character size in pixels	Width	4	5	6	7	8	8	10	12	16	16
	Height	6	12	8	12	8	12	18	16	8	12
Glyph magnified 2x

Eighty-one of the characters in table one form an orthogonal set of box drawing characters, here called the plain-heavy system. Orthogonal means that each of the four arms can be selected separately to be absent, plain, or heavy. A space character was required to complete the chart; chosen was character 0020 from many options.

This and later tables often contain a gif image and hexadecimal code only.

Table two.
	Col 1	Col 2	Col 3	Col 4	Col 5	Col 6	Col 7	Col 8	Col 9
Row 1	0020	2574	2578	2577	2510	2511	257B	2512	2513
Row 2	2576	2500	257E	250C	252C	252D	250E	2530	2531
Row 3	257A	257C	2501	250D	252E	252F	250F	2532	2533
Row 4	2575	2518	2519	2502	2524	2525	257D	2527	252A
Row 5	2514	2534	2535	251C	253C	253D	251F	2541	2545
Row 6	2515	2536	2537	251D	253E	253F	2522	2546	2548
Row 7	2579	251A	251B	257F	2526	2529	2503	2528	252B
Row 8	2516	2538	2539	251E	2540	2543	2520	2542	2549
Row 9	2517	253A	253B	2521	2544	2547	2523	254A	254B

Directory of table two.
	Absent	Plain	Heavy
North arm	Rows 1, 2, 3	Rows 4, 5, 6	Rows 7, 8, 9
East arm	Rows 1, 4, 7	Rows 2, 5, 8	Rows 3, 6, 9
South arm	Columns 1, 2, 3	Columns 4, 5, 6	Columns 7, 8, 9
West arm	Columns 1, 4, 7	Columns 2, 5, 8	Columns 3, 6, 9

The 40 box drawing characters of code page 437 are retained, but the system is not expanded into orthogonality. The sample glyphs in this report render a plain stroke of this plain-double system no different from the plain stroke of the plain-heavy system. Below each image are its Unicode number (in hexadecimal) and its number (in decimal) from code page 437.

Table three.
250C 218	252C 194	2510 191	2500 196	2553 214	2565 210	2556 183
251C 195	253C 197	2524 180		255F 199	256B 215	2562 182
2514 192	2534 193	2518 217		2559 211	2568 208	255C 189
2502 179			0020 020	2551 186
2552 213	2564 209	2555 184	2550 205	2554 201	2566 203	2557 187
255E 198	256A 216	2561 181		2560 204	256C 206	2563 185
2558 212	2567 207	255B 190		255A 200	2569 202	255D 188

A number preceded by the letter 'N' identifies a character that is not part of Unicode, but included in this report for purposes of comparison.

Some characters in the plain-heavy system	2578	257F	2527	2532	253D	2545	2549
Equivalent plain-double characters not in Unicode	N-11	N-12	N-13	N-14	N-15	N-16	N-17

Unicode also does not have box-drawing characters that mix heavy arms and double arms:

N-18

N-19

Meanwhile, the curved figures offer neither a heavy nor a double option:

256D

256E

256F

2570

Characters 2571, 2572 and 2573 involve diagonals.

2571

2572

2573

As with the curved characters, there are no double strokes nor heavy strokes. Many combinations of arms do not appear:

N-21

N-22

N-23

N-24

Strokes of the diagonal characters meet the edges of the bounding box at the corners, not the midpoints of the sides. In this regard, they are incompatible with the other 125 box-drawing characters, as strokes will not connect. This could have been addressed by making the diagonal characters similar to the examples below, which resemble characters 44 through 4B (hexadecimal) of Videotex:

N-25

N-26

N-27

N-28

N-29

In this report, glyphs for the dashed lines are asymmetric; one reason is that few displays will have enough pixels to allow all the dashed figures to be symmetric.

254C	2504	2508	254E	2506	250A
254D	2505	2509	254F	2507	250B

Although plain and heavy lines are supported, double lines are not.

N-31

N-32

N-33

N-34

N-35

N-36

Block elements 2581-258F suggest that the definers of Unicode had in mind that a character cell would be eight pixels high and eight pixels wide, or some multiple of eight, as illustrated by these examples:

Character	2581	2582	2583	2589	258A	258B
Unicode Name	lower one-eighth block	lower one-quarter block	lower three-eighths block	left seven-eighths block	left three-quarters block	left five-eighths block

The Unicode characters allow drawing at the bottom or left-hand side of the character position. To draw on the top or right, inverse video is required:

Normal video	2581	2582	2583	2584	2585	2586	2587
Inverse video	2581	2582	2583	2584	2585	2586	2587
Normal video				2580			2594

Normal video	2589	258A	258B	258C	258D	258E	258F
Inverse video	2589	258A	258B	258C	258D	258E	258F
Normal video	2595			2590

Within each character, the drawn and undrawn portions are contiguous respectively, hence characters such as these are not included:

Normal video	N-41	N-42	N-43	N-44
Inverse video	N-45	N-46	N-47	N-48

Retained from code page 437 are the shaded characters, each of which consumes the entire area of the cell:

Normal video	0020	2591	2592	2593	2588
Inverse video	0020	2591	2592	2593	2594
Normal video	2588	2593	2592	2591	0020

At the time that the Unicode standard was developing, inverse video was available on some, but not all, displays. When inverse video is available, characters 2580, 2588, 2590, 2593 and 2594 are redundant. When on the other hand inverse video is not available, then an irregular subset of the block element characters is unsupported. Explaining the inconsistency is that Unicode incorporates some older standards in toto, even if those standards are not entirely compatible.

Note also that 259A the invere of 259E, 2598 is the inverse of 259F, et cetera.

259A	2598	259D	2597	2596
259E	259F	2599	259B	259C

Unicode has a wealth of other symbols similar to those discussed so far, including the terminal graphics characters:

23B8 ⎸

2502 │

23B9 ⎹

23BA ⎺

23BB ⎻

2500 ─

23BC ⎼

23BD ⎽

In some renderings, the glyph of 23B8 might not differ from that of 258F, 23BD from 2581, et cetera.

Certain of the Unicode geometric shapes are shown below. As drawn, each triangle spans the full width of the cell, but not the full height. The standard is silent on whether a narrower or taller triangle is acceptable.

25E2 ◢

25E3 ◣

25E4 ◤

25E5 ◥

Something that would combine well with the box elements is a tall triangle:

25E2

25E3

25E4

25E5