Quite late in my research on magic cubes, I happily came
into possession of copies of Rev. A. H. Frost's early papers on magic cubes.
These papers quickly upset many of my preconceptions regarding magic cubes. The
earliest previous publication of cubes (that I am aware of) that were magic by
present standards (rows, columns, pillars, and triagonals all correct) was an
order 3 cube by Hugel in 1876 and order
4 and 5 cubes published in 1899 by Schubert.
The first publication of a
perfect magic cube was by Ian Howard in a JRM paper in 1976 (he gave
instructions on building an order 11).
ADDENDUM: F. Barnard published a lengthy paper in 1888 which included 3 perfect
magic cubes and mentioned the .Frankenstein cube (not perfect) of 1875. Arnoux
constructed an order 17 perfect cube in 1887, but his paper was not rediscovered
until 2003! [6] [7]
A well-researched and written Biography of Dr. Frost by Christian Boyer is on his Multimagic pages.
On this page I will present all of Frost’s cubes. They will
appear in the same order and the same orientation, exactly as they are in his
papers.
Following are the references to the papers I have been referring to. There must
be a preceding paper, which I do not have. In [3] page 49, Frost says “By the
method adopted in No. XXV, 1865, of this journal…” , and again in [4] page 118, he
says “The method employed for squares of the form 4n in No. 25, 1865, of
this Journal…”. That paper presumably deals with nasik magic squares, but I have
not been able to locate it.
About the term Nasik
Dr. Frost coined the term Nasik for a magic square in which both diagonals and all parallel diagonal pairs sum to the magic constant. This type of magic square would later be referred to as Pandiagonal or Perfect. In fact it is a Perfect Magic Hypercube of dimension 2. Later, Dr. C. Planck would extend the definition of Nasik to Perfect Magic Hypercubes of any dimension. [8]
The term Nasik is derived from the town Nassick, in Western India , where Frost was stationed from 1855 to 1867 as a missionary.
![]()
[1] A. H. Frost, Invention of Magic Cubes. Quarterly
Journal of Mathematics, 7, 1866, pp 92-103
He describes a method of constructing magic cubes and shows an order 7
pandiagonal and an order 8 pantriagonal magic cube. Frost was presumably still
in India at this time, because the paper was submitted by his brother, Rev.
Percival Frost
[2] A. H. Frost, Supplementary Note on Magic Cubes.
Quarterly Journal of Mathematics, 8, 1867, p 74
A short description of a coordinated set of 7 cubes, details and illustrations
of which were published in [4] and [5]
[3] A. H. Frost, On the General Properties of Nasik
Squares, QJM 15, 1878, pp 34-49
Construction of pandiagonal magic squares. This is the origin of the term
nasik, for what later become popularly known as pandiagonal magic
squares.
[4] A. H. Frost, On the General Properties of Nasik
Cubes, QJM 15, 1878, pp 93-123 plus plates 1 and 2.
He shows two order 3 and order 4 cubes, and one each of orders 7 and 9, with
method of construction. These cubes (in order) are not magic, disguised order 3,
not magic, pantriagonal, pantriagonal and perfect.
He briefly discusses order 5 but shows no complete order 5 cube.
[5] A. H. Frost, Description of Plates 3 to 9,
QJM
15, 1878, pp 366-368 plus plates 3 to 9.
Illustrations of a group of 7 interrelated order 7 cubes.
All the cubes shown on this Web page appear in either [1] or [4].
[6] F.A.P. Barnard,
Theory of Magic Squares and Magic Cubes, Memoirs of the National
Academy of Science, 4,1888, pp. 207-270.
A lengthy footnote describes the construction of G. Frankenstein's diagonal type
order 8 cube of 1875.
[7] See my pages on Arnoux.
[8] The Theory of Path Nasiks, by C. Planck, M.A., M.R.C.S., Private printing, 1905.
![]()
Order 7 Pandiagonal 1866 |
22 pandiagonal and 5 simple magic squares |
Order 8 Pantriagonal 1866 |
no magic squares. Corners of all order 3, 5, and 7 cubes sum correctly |
Order 3 Not magic 1878 |
rows and columns o.k. but only 2 triagonals are correct |
Order 3 Associated 1878 |
disguised version of No 2 basic cube. |
Order 4 Not magic 1878 |
4 pandiagonal magic squares. Triagonals are incorrect. |
Order 4 Pantriagonal 1878 |
no special features except compact. |
Order 7 Pandiagonal 1878 |
21 pandiagonal and 6 simple magic squares |
Order 9 Perfect 1878 |
33 pandiagonal magic squares (but not consecutive numbers) |
A presentation to Rev. Frost |
Picture of a silver cup he received from native friends in India. |
![]()
This pandiagonal associated magic cube contains 22 pandiagonal and 5 simple
magic squares.
It is almost certainly the first magic cube of modern definition to be
published. To be as large as order 7 and as feature rich is astounding! Until W.
S. Andrews defined the minimum requirements for a magic cube 42 years later,
many such cubes did not have the triagonals (space diagonals) summing correctly.
In fact, many did not even have the rows, columns, or pillars summing correctly,
but were concerned only with the total for each complete planar array.
This is a pandiagonal magic cube, so all 21 planar squares are pandiagonal magic. In addition 1 oblique square is also pandiagonal magic and the other 5 are simple magic. Four of the 5 simple magic squares have all pandiagonals in one direction summing correctly to 1204.
It is interesting that Frost calls this type of cube Nasik , which we all know is another term (coined by him) for pandiagonal magic squares. John Hendricks was unaware of Frost’s nasik cubes when he named this type of magic cube pandiagonal.
Frost also referred to the order 8 cube (following) in this first paper, as nasik.
The order 7 cube is nasik (pandiagonal) magic because all the planar squares
are nasik magic. We now call Frost’s second cube (the order 8), pantriagonal
magic because all pantriagonals sum correctly. Frost presumably classed it with
the order 7, because all pandiagonals of the 6 oblique squares sum
correctly. In a pandiagonal cube, it seems these 6 oblique squares will always
be magic, in some combination of pandiagonal and simple. In a
pantriagonal magic cube, it seems these 6 oblique squares are never magic,
because rows and columns do not sum correctly!
As mentioned above, all 4 triagonals sum correct. All lines parallel to 2 of these triagonals (the broken triagonals) are correct for 2 of the 4 directions.
[1] A. H. Frost, Invention of Magic Cubes. Quarterly Journal of Mathematics, 7, 1866, page 100
Top II 327 41 98 99 156 213 270 113 170 227 284 341 6 63 52 109 166 223 280 330 44 237 294 295 9 66 123 180 169 226 283 340 5 62 119 305 19 76 133 183 240 248 293 301 8 65 122 179 236 79 136 193 201 258 315 22 18 75 132 189 239 247 304 154 204 261 318 32 89 146 135 192 200 257 314 28 78 271 328 42 92 100 157 214 210 260 317 31 88 145 153 45 53 110 167 224 274 331 III IV 249 306 20 77 127 184 241 91 141 149 206 263 320 34 23 80 137 194 202 259 309 159 216 273 323 37 94 102 147 148 205 262 319 33 90 276 333 47 55 112 162 219 215 272 329 36 93 101 158 1 58 115 172 229 286 343 332 46 54 111 168 218 275 125 182 232 289 297 11 68 57 114 171 228 285 342 7 242 250 307 21 71 128 185 181 238 288 296 10 67 124 310 24 81 138 195 203 253 V VI 220 277 334 48 56 106 163 13 70 120 177 234 291 299 337 2 59 116 173 230 287 130 187 244 252 302 16 73 69 126 176 233 290 298 12 198 255 312 26 83 140 190 186 243 251 308 15 72 129 322 29 86 143 151 208 265 254 311 25 82 139 196 197 96 104 161 211 268 325 39 35 85 142 150 207 264 321 164 221 278 335 49 50 107 103 160 217 267 324 38 95 281 338 3 60 117 174 231 VII 191 199 256 313 27 84 134 266 316 30 87 144 152 209 40 97 105 155 212 269 326 108 165 222 279 336 43 51 225 282 339 4 61 118 175 300 14 64 121 178 235 292 74 131 188 245 246 303 17
This is the second of two cubes Frost presented in his paper on the Invention of Magic Cubes. The first cube (shown above) he showed was an order 7 pandiagonal magic cube. These two cubes are in QJM 7, 1866, pp 100, 101
This order 8 cube is called a pantriagonal magic cube because all pantriagonals sum correctly. This type of cube is possible starting with order 4! Frost also called it a Nasik (pandiagonal) magic cube, probably because the diagonals of the oblique squares are correct.
Pantriagonal magic cubes have a feature similar to pandiagonal magic squares:
Note that this particular pantriagonal magic cube has a bonus feature. All 8 corner cells of all cubical arrays within this cube that have sides of 3, 5 or 7 sum correctly to the constant. This includes wrapping around from 1 edge to the opposite edge. Corners of many other parrallelopiped arrays also sum correctly. This feature is fairly common with order 8 magic cubes and for perfect cubes of this order, it seems that all orders of the sub-cubes sum correctly.
This cube contains no magic squares. The 3 x 8 planar squares have all rows and columns correct. The diagonals of these squares are incorrect, but the 2 diagonals of each square are equal value and in 2 of the 3 cases form a pattern.
|
Diagonal sums of each planar square |
||
|
Horizontal |
Vertical |
Vertical
|
|
1796 |
1540 |
3028 |
|
2308 |
2564 |
1044 |
|
2308 |
2564 |
1012 |
|
1796 |
1540 |
3124 |
|
2308 |
2564 |
1076 |
|
1796 |
1540 |
3060 |
|
1796 |
1540 |
3092 |
|
2308 |
2564 |
980 |
This cube, as well as being pantriagonal magic, is called complete, because each pantriagonal contains 4 compliment pairs, with the members of each pair spaced 1/2m apart on the triagonal.
Top II 257 255 254 260 385 127 126 388 240 274 275 237 112 402 403 109 252 262 263 249 124 390 391 121 277 235 234 280 405 107 106 408 248 266 267 245 120 394 395 117 281 231 230 284 409 103 102 412 269 243 242 272 397 115 114 400 228 286 287 225 100 414 415 97 64 450 451 61 192 322 323 189 465 47 46 468 337 175 174 340 453 59 58 456 325 187 186 328 44 470 471 41 172 342 343 169 457 55 54 460 329 183 182 332 40 474 475 37 168 346 347 165 52 462 463 49 180 334 335 177 477 35 34 480 349 163 162 352 III IV 224 290 291 221 96 418 419 93 305 207 206 308 433 79 78 436 293 219 218 296 421 91 90 424 204 310 311 201 76 438 439 73 297 215 214 300 425 87 86 428 200 314 315 197 72 442 443 69 212 302 303 209 84 430 431 81 317 195 194 320 445 67 66 448 481 31 30 484 353 159 158 356 16 498 499 13 144 370 371 141 28 486 487 25 156 358 359 153 501 11 10 504 373 139 138 376 24 490 491 21 152 362 363 149 505 7 6 508 377 135 134 380 493 19 18 496 365 147 146 368 4 510 511 1 132 382 383 129 V VI 321 191 190 324 449 63 62 452 176 338 339 173 48 466 467 45 188 326 327 185 60 454 455 57 341 171 170 344 469 43 42 472 184 330 331 181 56 458 459 53 345 167 166 348 473 39 38 476 333 179 178 336 461 51 50 464 164 350 351 161 36 478 479 33 128 386 387 125 256 258 259 253 401 111 110 404 273 239 238 276 389 123 122 392 261 251 250 264 108 406 407 105 236 278 279 233 393 119 118 396 265 247 246 268 104 410 411 101 232 282 283 229 116 398 399 113 244 270 271 241 413 99 98 416 285 227 226 288 VII VIII 160 354 355 157 32 482 483 29 369 143 142 372 497 15 14 500 357 155 154 360 485 27 26 488 140 374 375 137 12 502 503 9 361 151 150 364 489 23 22 492 136 378 379 133 8 506 507 5 148 366 367 145 20 494 495 17 381 131 130 384 509 3 2 512 417 95 94 420 289 223 222 292 80 434 435 77 208 306 307 205 92 422 423 89 220 294 295 217 437 75 74 440 309 203 202 312 88 426 427 85 216 298 299 213 441 71 70 444 313 199 198 316 429 83 82 432 301 211 210 304 68 446 447 65 196 318 319 193
![]()
![]() |
Rows, columns and pillars all sum correctly on all 12 planar squares so the square sums to 3 x 42 = 126. Two oblique squares have all rows summing correctly and 4 have all columns correct. Only 2 of the 4 triagonals are correct but one of these has all pantriagonals correct. It is in QJM 15, 1878, page 97
|
![]() |
This cube has the exact same features as the preceding
one, except all 4 triagonals are correct, making it a true magic cube. There are a total of 4 basic order 3 magic cubes.
|
![]()
I obtained this cube originally from Christian Boyer (France) via email. He tipped me off as to its origin, resulting in my eventually obtaining copies of these Frost papers. It is in QJM 15, 1878, pp 116-117.
Horizontal planes are pandiagonal magic. Rows and columns of the other planes are correct. However, only 1 triagonal is correct so by present standards this is not a magic cube.
![]() |
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
![]()
This cube is pantriagonal magic, non-associated. It is in QJM 15,
1878, page 119.
It contains no magic squares, but because each triagonal (and broken triagonal)
contains two complement pairs with the members of each air spaced 1/2m
apart, the cube is called complete.
![]() |
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Frost presents
a simple method I have not seen before, for constructing the above cube.
Put successive odd numbers from 1 to 63 in the cells of the planes as
arranged here.
|
1 |
-3 |
-5 |
7 |
|
-17 |
19 |
21 |
-23 |
|
-33 |
35 |
37 |
-39 |
|
49 |
-51 |
-53 |
55 |
|
-9 |
11 |
13 |
-15 |
|
25 |
-27 |
-29 |
31 |
|
41 |
-43 |
-45 |
47 |
|
-57 |
59 |
61 |
-63 |
|
-37 |
39 |
33 |
-35 |
|
53 |
-55 |
-49 |
51 |
|
5 |
-7 |
-1 |
3 |
|
-21 |
23 |
17 |
-19 |
|
45 |
-47 |
-41 |
43 |
|
-61 |
63 |
57 |
-59 |
|
-13 |
15 |
9 |
-11 |
|
29 |
-31 |
-25 |
27 |
Add 65 to each number and divide by 2 to get the final numbers for the finished cube.
This magic cube pandiagonal and associated. It contains 21 pandiagonal (the planar) and 6 simple (the oblique) magic squares. It is in QJM 15, 1878, pp 99-101
Top II 205 146 31 266 151 85 320 42 270 155 96 324 216 101 262 154 88 316 208 142 34 92 327 212 104 38 273 158 319 204 145 30 265 150 91 100 41 269 161 95 323 215 33 261 153 87 322 207 141 157 98 326 211 103 37 272 90 318 210 144 29 264 149 214 99 40 268 160 94 329 147 32 260 152 86 321 206 271 156 97 325 217 102 36 148 89 317 209 143 35 263 328 213 105 39 267 159 93 III IV 166 51 335 220 112 46 274 339 231 116 1 285 170 62 223 108 49 277 162 54 331 4 281 173 58 342 227 119 280 165 50 334 219 111 45 61 338 230 115 7 284 169 330 222 107 48 276 168 53 118 3 287 172 57 341 226 44 279 164 56 333 218 110 175 60 337 229 114 6 283 52 336 221 106 47 275 167 225 117 2 286 171 63 340 109 43 278 163 55 332 224 282 174 59 343 228 113 5 V VI 120 12 289 181 66 301 235 251 185 77 305 239 131 16 177 69 297 238 123 8 292 308 242 127 19 247 188 73 234 126 11 288 180 65 300 15 250 184 76 304 245 130 291 176 68 296 237 122 14 72 307 241 133 18 246 187 299 233 125 10 294 179 64 129 21 249 183 75 303 244 13 290 182 67 295 236 121 186 71 306 240 132 17 252 70 298 232 124 9 293 178 243 128 20 248 189 74 302 VII 81 309 201 135 27 255 196 138 23 258 192 84 312 197 195 80 315 200 134 26 254 203 137 22 257 191 83 311 253 194 79 314 199 140 25 310 202 136 28 256 190 82 24 259 193 78 313 198 139
This must be the first perfect magic cube ever published. This at a time when others were producing magic(?) cubes with incorrect diagonals (and in some cases incorrect rows and columns.
In 1866 Frost published an order 7 pandiagonal and an order 8 pantriagonal magic cube. This cube is perfect because it combines the features of both the pandiagonal and the pantriagonal cubes. The 3 x 9 planar squares are pandiagonal magic (pandiagonal cube), as are the 6 oblique squares. All 4m2 pantriagonals sum correctly (pantriagonal cube). Total summations to the constant 4005 is m2 rows, m2 columns, m2 pillars, 6m2 pandiagonals, and 4m2 pantriagonals. Total summations 13m2 . The diagonal magic cube which was originally called perfect, has 3m2 + 6m + 4 correct lines!
However, this cube is not normal, because it uses numbers
from 1 to 889 instead of the consecutive numbers from 1 to 729. Frost points out
[1, p 115] that the smallest order cube possible with consecutive numbers, using
this method is order 11. Ian Howard mentioned the same thing when he published
[2] his method for constructing the order 11 perfect magic cube 98 years later!
We now know that order 8 and order 9 normal perfect magic cubes do exist,
presumably constructed using other methods.
This cube is also associated. Pairs of numbers
diametrically equidistant to the center number of 445 sum to 2 times that
number. Note also that 445 is one-half the sum of the first and last numbers
used in the cube.
Note further, that because the cube is associated, the 3 central orthogonal
magic squares are also associated. This feature is common to associated magic
hypercubes of any dimension. The lesser central hypercubes will always be magic
and associated.
[1] Quarterly Journal of
Mathematics 15 1878 pp 110-116
[2] Ian P. Howard, Pan-diagonal Associative Magic Cubes (Letter to the
Editor), JRM 9:4, 1976, pp276-278.
Top II 824 606 733 149 258 85 362 471 517 55 382 461 577 814 626 703 139 248 473 519 828 605 732 141 257 84 366 131 247 54 386 463 579 818 625 702 88 365 472 511 827 604 736 143 259 624 706 133 249 58 385 462 571 817 142 251 87 364 476 513 829 608 735 573 819 628 705 132 241 57 384 466 607 734 146 253 89 368 475 512 821 388 465 572 811 627 704 136 243 59 516 823 609 738 145 252 81 367 474 242 51 387 464 576 813 629 708 135 369 478 515 822 601 737 144 256 83 707 134 246 53 389 468 575 812 621 255 82 361 477 514 826 603 739 148 816 623 709 138 245 52 381 467 574 731 147 254 86 363 479 518 825 602 469 578 815 622 701 137 244 56 383 III IV 616 723 109 238 45 352 481 567 874 342 451 587 864 676 713 129 208 35 569 878 615 722 101 237 44 356 483 207 34 346 453 589 868 675 712 121 355 482 561 877 614 726 103 239 48 716 123 209 38 345 452 581 867 674 231 47 354 486 563 879 618 725 102 869 678 715 122 201 37 344 456 583 724 106 233 49 358 485 562 871 617 455 582 861 677 714 126 203 39 348 873 619 728 105 232 41 357 484 566 31 347 454 586 863 679 718 125 202 488 565 872 611 727 104 236 43 359 124 206 33 349 458 585 862 671 717 42 351 487 564 876 613 729 108 235 673 719 128 205 32 341 457 584 866 107 234 46 353 489 568 875 612 721 588 865 672 711 127 204 36 343 459 V VI 773 119 228 5 332 441 557 884 666 431 547 854 686 763 179 218 25 302 888 665 772 111 227 4 336 443 559 24 306 433 549 858 685 762 171 217 442 551 887 664 776 113 229 8 335 173 219 28 305 432 541 857 684 766 7 334 446 553 889 668 775 112 221 688 765 172 211 27 304 436 543 859 116 223 9 338 445 552 881 667 774 542 851 687 764 176 213 29 308 435 669 778 115 222 1 337 444 556 883 307 434 546 853 689 768 175 212 21 555 882 661 777 114 226 3 339 448 216 23 309 438 545 852 681 767 174 331 447 554 886 663 779 118 225 2 769 178 215 22 301 437 544 856 683 224 6 333 449 558 885 662 771 117 855 682 761 177 214 26 303 439 548 VII VIII 169 278 15 322 401 537 844 656 783 507 834 646 753 189 268 75 312 421 655 782 161 277 14 326 403 539 848 316 423 509 838 645 752 181 267 74 531 847 654 786 163 279 18 325 402 269 78 315 422 501 837 644 756 183 324 406 533 849 658 785 162 271 17 755 182 261 77 314 426 503 839 648 273 19 328 405 532 841 657 784 166 831 647 754 186 263 79 318 425 502 788 165 272 11 327 404 536 843 659 424 506 833 649 758 185 262 71 317 842 651 787 164 276 13 329 408 535 73 319 428 505 832 641 757 184 266 407 534 846 653 789 168 275 12 321 188 265 72 311 427 504 836 643 759 16 323 409 538 845 652 781 167 274 642 751 187 264 76 313 429 508 835 IX 288 65 372 411 527 804 636 743 159 742 151 287 64 376 413 529 808 635 807 634 746 153 289 68 375 412 521 416 523 809 638 745 152 281 67 374 69 378 415 522 801 637 744 156 283 155 282 61 377 414 526 803 639 748 631 747 154 286 63 379 418 525 802 524 806 633 749 158 285 62 371 417 373 419 528 805 632 741 157 284 66
![]()
![]()
The Reverend Andrew Hollingworth Frost (1820-1907), a mathematics wrangler of St. John's College, Cambridge, was a church missionary in Nasik, Bombay, India, from 1855 to 1867. [1], [2]
[1] Ollerenshaw and
Brée, Most-Perfect
Pandiagonal Magic Squares, IMA, 1998, 0-905091-06-X,
footnote on page 6.
[2] A well-researched and written Biography of Dr. Frost by
Christian Boyer is on his Multimagic pages.
![]()

Please send me Feedback about my Web
site!![]()
![]()
![]()
Harvey Heinz harveyheinz@shaw.ca
This page last updated
October 14, 2009
Copyright © 2003 by Harvey D. Heinz