Ο                                   ΟΟ
ΟΟΟ                            ΟΟΟΟ
ΟΟΟΟΟ                      ΟΟΟΟΟΟ
ΟΟΟΟΟΟΟ                ΟΟΟΟΟΟΟΟ

Animals and so-called primitive peoples do not bother to make nice distinctions between entities on the basis of number and even today, when  deprived of technological aids, we are not at all good at it (Note 1).  What people do ‘naturally’ is to make distinctions of type not number and the favourite principle of division by type is the two-valued either/or principle.  Plato thought that this principle, dichotomy, was so fundamental that all knowledge was based on it — the reason for this being because the brain works in this way, the nerve synapsis is either ‘on’ or ‘off’. Psychologically human beings have a very strong inclination to proceed by straight two-valued distinctions, light/dark, this/that, on/off, sacred/profane, Greek/Barbarian, Jew/Gentile, good/evil and so on — more complex gradations are only introduced later and usually with great reluctance Science has eventually recognized the complexity of nature and apart from gender there are not many true scientific dichotomies left though we still have the classification of animals into  vertebrates and invertebrates.

Numbers themselves very early on got classified into even and odd , the most fundamental numerical distinction after the classification one and many which is even more basic.

The classification even/odd is radical: it provides what modern mathematicians call a partition of the whole set. That is, the classification principle is exhaustive : with the possible exception of the unit, all  numbers fall into one or other of the two categories. Moreover, the two classes are mutually exclusive: no number appearing in the list of evens will appear in thelist of odds. This is by no means true of all classification principles for numbers as one might perhaps at first assume. Numbers can be classified, for example, as triangular and as rectangular according to whether they can be (literally) made into rectangles or equilateral triangles. But ΟΟΟΟΟΟ turns out to be both since it can be formed either into a triangle or a rectangle:

ΟΟΟ                                         ΟΟΟ
ΟΟΟ                                         ΟΟ
Ο
The Greeks, like practically all cultures in the ancient world, viewed the odd and even numbers as male and female respectively — presumably because a woman has ‘two’ breasts and a male only one penis. And, since oddness, though in Greek the term did not have the same associations as in English, was nonetheless defined with respect to evenness and not the reverse, this made an odd number a sort of female manqué. This must have posed a problem for their strongly patriarchal society but the Greek philosophers and mathematicians got round this by arguing that ‘one’  (and not ‘two’) was the basis of the number system while ‘one’ was the ‘father of all numbers’.

On the other hand a matriarchal society or a species where females were dominant would almost certainly, and with better reasoning, have made ‘one’ a female number, the primeval egg from which the whole numerical progeny emerged. Those who consider that mathematics is in some sense ‘eternally true’ should reflect on the question of how mathematics would  have developed within a hermaphroditic species, or in a world where there were three and not two humanoid genders as in Ian Banks’s science-fiction novel  The Player of Games.

Evenness is not easy to define — nor for that matter to recognize as I have just realized since, coming across an earlier version of this section, I found I was momentarily incapable of deciding which of the rows of balls pictured at the head of this chapter represented odd or even numbers. We have to appeal to some very basic feeling for ‘symmetry’ — what is on one side of a dividing line is exactly matched by what is on the other side of it. A definition could thus be

If you can pair off a collection within itself and nothing remains over, then the collection is called even, if you cannot do this the collection is termed odd.

This makes oddness anomalous and less basic than evenness which intuitively one feels to be right —  we would not, I think, ever dream of defining oddness and then say “If a collection is not odd, it is even”. And although it is only in English and a few other languages that ‘odd’ also means ‘strange’, the pejorative sense that the word odd has picked up suggests that we expect and desire things to match up, i.e. we expect, or at least desire, them  to be ‘even’ —  the figure of Justice holds a pair of evenly balanced scales.

The sense of even as ‘level’ may well be the original one. If we have two collections of objects which, individually,  are more or less identical, then a pair of scales remains level if the collections are placed on each arm of the lever (at the same distance).  One could define even and odd thus pragmatically:

“If a collection of identical standard objects can be divided up in a way which keeps the arms of a balance level, then the collection is termed even. If this is not possible it is termed odd.”

This definition avoids using the word two which is preferable since the sense of things being ‘even’ is much more fundamental than a feeling for ‘twoness’  — for this reason the distinction even/odd, like the even more fundamental ‘one/many’ , belongs to the stage of pre-numbering rather than that of numbering.

Early man would not have had a pair of scales, of course, but he would have been familiar with the procedure of ‘equal division’, and the simplest way of dividing up a collection of objects is to separate it into two equal parts. If there was an item left over it could simply be thrown away. Evenness is thus not only the simplest way of dividing up a set of objects but the principle of division which makes the remainder a minimum: any other method of division  runs the risk of having more objects left over.

Euclid’s definition is that of equal division. He says “An even number is that which is divisible into two equal parts” (Elements Definition 6. Book VII)  and “An odd number is that which is not divisible into two equal parts, or that which differs by  a unit from an even number”  (Elements  Definition 7. Book VII). Incidentally, in Euclid ‘number’ not only always has the sense ‘positive integer but has a concrete sense — he defines ‘number‘ as a “multitude composed of units”.

Note that Euclid defines odd first privatively (by what it is not) and then as something deficient with reference to an even number. The second definition is still with us today: algebraically the formula for the odd numbers is (2n-1) where n is given the successive values 1, 2, 3…. or sometimes (in order to leave 1 out of it) by giving n the successive values 2, 3, 4….  In concrete terms,  we have the sequence

Ο     ΟΟ     ΟΟΟ  ……..                 …..

Duplicating them gives us the ‘doubles’ or even numbers

Ο     ΟΟ     ΟΟΟ  ..….
Ο     ΟΟ     ΟΟΟ  ……

and  removing a unit each time gives us the ‘deficient’ odd numbers.

The unit itself is something out on its own and was traditionally regarded as  neither even nor odd. It is certainly not even according to the ‘equal division’ definition since it cannot be divided at all (within the context of whole number theory) and it cannot be put on the scales without disturbing equilibrium. In practice it is often convenient to treat the unit as if it were odd, just as it is to consider it a square number, cube number and so forth, otherwise many theorems would have to be stated twice over. Context usually makes it clear whether the term ‘number’ includes the unit or not.

Note that distinguishing between even and odd has nothing to do with counting or even with distinguishing between greater or less – knowing that a number is even tells you nothing about its size. And vice-versa, associating a number word or symbol with a collection of objects will not inform you as to  whether the quantity is even or odd — there are no ‘even’ or ‘odd’ endings to the spoken word like those showing whether something is singular or plural,  masculine or feminine.

It is significant that we do not have words for numbers which, for example, are multiples of four or which leave a remainder of one unit when divided into three. (The Greek mathematicians did, however, speak of ‘even-even’ numbers.) If our species had three genders instead of two, as in the world described in The Player of Games, we would maybe tend to divide things into threes and classify all numbers according to whether they could be divided into three parts exactly, were a counter short or a counter over. This, however, would have made things so much more complicated that such a species would most likely have taken even longer to develop numbering and arithmetic than in our own case.

The distinction even/odd is the first and simplest case of what is today called a congruence. The integers can be separated out into so-called equivalence classes according to the remainder left when they are divided by a given number termed the modulus. All numbers are in the same class (modulus 1) since when they are separated out into ones there is only one possible remainder : nothing at all. In Gauss’s notation the even numbers are the numbers which leave a remainder of zero when divided by 2, or are ‘0 (mod 2)’ where mod is short for modulus. And the odd numbers are all 1 (mod 2) i.e. leave a unit when separated into twos. What is striking is that although the distinction between even and odd, i.e. distinction between numbers that are 0 or 1 (mod 2) is prehistoric, congruence arithmetic as such was invented by Gauss a mere couple of centuries ago.

In concrete terms we can set up equivalence classes relative to a given modulus by arranging collections of counters (in fact or in imagination) between parallel lines of set width starting with unit width, then a width which allows two counters only, then three and so on. This image enables us to see at once that the sum of any two or more even numbers is always even.

And since an odd number has an extra  Ο  this means a pair of odd numbers have each an extra unit and so, if we fit them together to make the units face each other we have an even result. Thus    Even plus even equals even” and “Odd plus odd equals even” are not just jingles we have to learn at school but correspond to what actually happens if we try to arrange actual counters or squares so that they match up.

We end up with the following two tables which may well have been the earliest ones ever to have been drawn up by mathematicians.

­­­­­­­­­­­­­­­­­­­­­­­          +       odd      even                        ×     odd    even  

       odd      even    odd                     odd    odd    even

       even    odd    even                     even  even  even

All this may seem so obvious that it is hardly worth stating but simply by appealing to these tables many results can be deduced that are far from being self-evident. For example, we find by experience that certain concrete  numbers can be arranged as rectangles and that, amongst these rectangular numbers, there are ones that can be separated into two smaller rectangles and those that cannot be. However if I am told that a certain collection can be arranged as a rectangle with one side just a unit greater than the other, then I can immediately deduce that it can be separated into two smaller rectangles. Why am I so sure of this? Because, referring to the tables above,

1.) the ‘product’ of an even and an odd number is even;
2.) an even number can by definition always be separated into two equal parts.

           I could deduce this even if I was a member of a society which had no written number system and no more than a handful of number words.

This is only the beginning: the banal distinction between even and odd and reference to the entries in the tables above crops up in a surprising amount of proofs in number theory. The famous proof that the square root of 2 is not a rational number — as we would put it — is based on the fact that no quantity made up of so many equal bits can be at once even and odd.                                                                       SH 5/03/15

Note 1  This fact (that human beings are not naturally very good at assessing numerical quantity) is paradoxical since mankind is the numerical animal par excellence. Mathematics is the classic case of the weakling who makes himself into Arnold Schwarzenegger. It is because we are so bad at quantitative assessment that playing cards are obliged to show the number words in the corner of the card and why the dots on a dice are arranged in set patterns to avoid confusion.

ADDITION , the basic arithmetic operation, is not quite so straightforward and unambiguous as one might suppose. When we ‘add’ one thing to another, or to a collection, the originally separate items remain separate after combination : they do not fuse or merge.

Addition is a strictly numerical operation which tells us nothing about the sizes of the objects that are brought together, nor their colour, mass and so forth. The only assumption made is that the items with which arithmetic  deals are discrete and remain so. Even when dealing with liquids which do combine together imperceptibly to form a whole, we often tend to still think in terms of discrete items : when we talk of ‘adding’ a bucketful of water to a pond, the implication is that the water in the pond is made up of ‘so many bucketfuls’, i.e. could be broken down into units. If we want to take merging into account we end up with the formula 1 + 1  = 1  which, though it is a perfectly correct representation of what goes on when, say, we combine two droplets of rainwater, looks extremely peculiar. It would be quite possible, though probably not worthwhile, to develop ‘Rainwater Arithmetic’ where no matter how many items you add together the net result is 1. Many liquids have an upper limit to merging : if you carry on adding droplets of oil to an initial droplet lying on water, the sheet of oil eventually splits in two. In such a case 1 + 100,000,000 = 2 or something of that order of magnitude. Our arithmetic, then, concerns entities with an upper limit of 1 , i.e. substances which never merge at all or, if they do , immediately break apart.

There are at least three different senses to addition which we might call  ‘adding’, ‘adding on’ and ‘adding up’. In the first case we join together two groups or collections of comparable size. In the second case we tack on a smaller collection to a larger, and in the third case we do not so much join together as rearrange. In abstract mathematics there is no difference in the operations involved but in concrete terms there is all the difference in the world.

The most ancient of the three types of addition is undoubtedly ‘adding on’. If we go back to the time when objects or events were recorded by notches on a bone or knots in a cord (which I shall call the tally system) there is nothing to ‘add up’ because there is no numerical base : what is on the stick or bone is the ‘total’ and that’s all there is to it.  A new item, the birth of a child or a caribou kill, will be recorded at the end of the list which will, if there are very few items most likely be arranged in a row, or in several rows more or less underneath each other as on the Ishango bone, perhaps the earliest example of written numerals. In Wild West days, Billy the Kid “had so many notches on his gun” — one notch, one dead person. The amount grows by accretion just like an organism : the longer the list the more items, children born, midsummers, caribou kills and so on.

But when you add ciphered numbers the final amount does not grow noticeably  : 2 + 3 = 5 where 5 is no larger or longer than 2 or 3. This is not in the least a trivial observation since it is the root cause of the current very widespread misunderstanding of what numbers essentially are, namely the symbolic representation of real or imaginary collections of certain objects, collections which increase or diminish in size.

And when you add up a column of figures all you are doing is getting the separate amounts into a tidier form. What was separate (distinct) before addition remains separate after addition : the difference is that the whole lot is now grouped in a systematic fashion according to the powers of the base, usually ten. But what is there at the end was there at the beginning.

However, when you ‘add’ fresh recruits to a regiment, new employees to a firm’s workforce or money to someone’s bank account you are not just rearranging what is already there but bringing in someone or something new from outside. It is at first sight somewhat surprising that this makes no difference arithmetically speaking, the reason being that by the time the new objects are actually joined to the existing group they are no longer completely ‘new’ since they are already in existence, even if, as in the case of an ‘addition’ to a family they have only just been born.

Despite the difference between adding up and adding on, the operation of addition remains, like division, an ‘inert’ one : strictly speaking nothing is created or destroyed. A primary schoolteacher teaching addition to a child has to have the extra building blocks required concealed to one side : if there is nothing there to add, addition cannot take place. This is quite distinct from a truly ‘creative’ operation where something new is produced from within as when a mother gives birth or when unicellular organisms like bacteria reproduce by splitting in two (mitosis). I did as a matter of fact start to  construct an arithmetical system where the basic operations are splitting and merging instead of adding and subtracting.   SH  14/9/13

Bases

The first thing to realize about bases is that Nature does not bother with them. Nature does not group objects into tens, hundreds, thousands and so on, not even into twos, fours, eights. Bases in the mathematical sense are entirely a matter of human convenience — Nature only uses base one. We are so used to thinking decimally and writing numbers in columns that we tend to consider that an amount, expressed in our modern Hindu-Arabic  positional numerals, is somehow truer than if it were expressed in, say, Chinese Stick Numbers. Even, there is a tendency to think that the modern representation of a quantity  is somehow numerically truer or more real  than the actual quantity  — an example of the delusionary thinking that doing mathematics all too readily gives rise to. Asked how many stones there are in a certain pile, an unhelpful but perfectly correct answer would be to transfer all the stones into a wheelbarrow and empty them out at the enquirer’s feet.

          So why do we bother with bases? Partly for reasons of space : an amount expressed in base one requires a lot of paper or wood or whatever material we are employing and this was and is an important consideration. But the main reason is that our perception of number is so defective that we find it very hard indeed  to distinguish between amounts of similar objects or marks beyond a certain very small quantity (seven at most). So what do we do? What we need is a second fixed quantity, a second ‘unit’, in terms of which we can assess larger quantities. What do we choose?

‘Ones’ are given us by Nature and the sort of objects we shall want to assess either are collections of ‘ones’, like trees or sheep, or can be treated as if they were, e.g. mountains, villages and so on. We speak of ‘whole’ numbers thus implying that they are in some sense ‘entire’, ‘indivisible’. If our standard ‘one-object’ is a pebble it really is indivisible in a pre-industrial society and a cowrie shell, though it can be broken in two, ceases to be a proper shell if this is done. There is, in concrete number systems,  no question of dividing a ‘one’ and ‘fractions’ (‘broken numbers’), if defined at all, are represented by smaller objects or marks, not by splitting up the ‘one-object’. Whole number arithmetic was by Greek times firmly associated with the physical theory of atomism as put forward by Democritus who himself wrote works on arithmetic now lost.

A mass of ‘ones’ is perceptually unmanageable unless related to certain standard amounts we are familiar with. But Nature is extremely unhelpful in providing us with exact standard amounts. Litters of kittens and puppies are by no means standard, and apart from a slight prejudice in favour of the quantity five, the amount of petals in a flower or branches on a tree varies amazingly.
Familiar fixed amounts such as the ‘number’ of hills on the skyline could only be of strictly local relevance and at the end of the day about the only available standard amounts given to man by Nature are the fingers which, counting the thumbs, come in fives. It is thus no surprise that number systems are dominated by the amounts five, ten and twenty.

Alternatively, of course, we could start from the other end and opt for a ‘man-made’ secondary unit whose size would depend on our perceptual needs and what exactly it is we want to assess or measure. These three criteria 1.) availability of a standard amount; 2.) human perceptual  limitations and 3.) appropriateness for assessment purposes, conflict and one of the main problems of early numbering was how to reach an acceptable compromise between them.

Two is the first possibility for a ‘secondary unit’ but, although it has come into its  own in the computer era (because of the two states On and Off), it is clearly too small to be of much use for ordinary  purposes. A language spoken in the Torres Straits had a word for our ‘one’, namely urapan and a word for our ‘two’ okasa and that was about it. Their numerals went

1.       urapan                   4.       okasa okasa

2.       okasa                     5.       okasa okasa urapan

3.       okasa urapan         

Understandably, since even a number as small as 11 would require six words, the natives referred to anything above 6 as ras — ‘a lot’ (Conant, p. 105).

A few things, or rather events, are viewed in threes witness phrases like “third time lucky” and we group quite a lot of things in fours (seasons, points of the compass &c.) but  the obvious first choice for a ‘secondary unit’ is five. Beyond five we really feel the need for a ‘secondary unit’ since  collections like    l l l l l l l l and   l l l l l l re  practically indistinguishable. Also, as it happens we have the five fingers to be able to check (by pairing off) whether we are separating out the items into groups correctly.

The Old Man of the Sea in Homer ‘fives’ his seals but for most herdsmen five would still have been rather too small as a secondary unit. So where do we go next? If we remain guided by the fingers the next possibilities are ‘both hands’ and what many primitive languages referred to as ‘the whole man’ i.e. ten and twenty.  Twenty is in some ways a better choice since, if we keep the option open of reverting to five for trifling amounts we can cope with very sizeable collections using batches of twenty. The Yoruba used twenty cowrie shells as their principal counting amount after the unit. Some modern European  languages which have long since become decimal show traces of an earlier vigesimal (twenty-based) system which probably suited farmers better. Hence Biblical terms like ‘three score years and ten’ in English and the French soixante dix-huit (sixty-eighteen).

A secondary  unit is, unlike the unit, not actually indivisible — since it is still made up of standard ones — so how do we keep it together if we are using objects as numbers? This depends on the choice of standard object and in practice is one of the motivations for the choice of object in the first place (or second place at least). Heaps of pebbles are heavy enough not to blow away but can all too easily be disturbed by people bumping into them, while piles of flat objects unless they are paper thin readily tip over and in any case really flat objects are hard to come by in nature. This is where shells are advantageous since if of the cowrie variety they stack up neatly and, even better, can be pierced and threaded on strings to make number rosaries. Beads make good numbers but since they are manufactured items they would not have been amongst the very earliest examples of object numbers.

The clay Number Ball I have already mentioned would not be suitable for secondary (or tertiary) standard amounts precisely because the bits tend to adhere together : its use would be for assessing limited quantities in the field which, if required, could be recorded back at the Number Hut using a different system, knotting or incising.

On my island I opt for a stick of standard length as my ‘one-object’ and I instruct the natives to tie sticks together into a bundle when we reach the fixed secondary amount which tentatively I fix at our twelve. Lacking a  sign on this computer for a bundle of sticks I represent this amount by ∏. At the back of the Number Hut I set up partitions to make alleyways for concrete numbers and I suspend from the roof an example of the bundle the alleyway is to contain and its decomposition into smaller bundles or individual sticks as a sort of Système Internationale prototype. For the moment I only propose to use the extreme right two alleyways, the first for individual sticks and the second for bundles of the specified size.  Whenever we have  ⁄ sticks in the extreme right alley, they are tied together and transferred (literally ‘carried’) to the next alleyway. The system can be used for the temporary recording of data but it is best to restrict its use to calculation, simple additions and subtractions, while using a different system for recording purposes. I can, for example, paint three vertical lines on a piece of bark to make it into a Number Board and paint in particular configurations of the sticks and bundles.  When painting I do not use short cuts, I just represent the sticks and bundles as well as I can turning stick numbers into stroke numbers.  As yet I do not proceed any further : all quantities are to be represented by sticks and/or Number Bundles of a single fixed amount and, for the time being, I do not set an upper limit to the amount of bundles. Thus the components of our number system so far are only  l  and  ∏   where

   ∏    =        l l l l l l l l l l l l  

A  feature of this still very rudimentary system is that at any moment a bundle   ∏  can be reduced to so many sticks simply by untying the cord and transferring them into the appropriate alley. Even, it is possible to have second thoughts about the ‘secondary unit’ and change it for another, since all you have to do is untie the bundles and tie the sticks up again using a different set amount. We might, for example, want to revert to five if the quantity to be assessed turns out not to be so great, or, conversely, jump ahead to twenty for a really large herd of goats or clump of trees. With systems that depend on threading objects on a string or wire, changing the secondary unit is either impossible or time-consuming and so would tend not to be done.

If the first ‘greater unit’ is set at ten and we are dealing with sticks, the problem of distinguishing different numbers less than ten  remains — we have met requirements 1.) and 3.) but not 2.) . The earliest Egyptian written numbers, perhaps based on still earlier number sticks, got round this problem, or tried to, by arranging the sticks in set patterns. But the patterns are not very distinctive or memorable. Far more striking are the excellent domino or dice dot numbers. Domino patterned numbers stop at six and the arrangements for the playing-card seven, eight and nine are not so striking — I have sometimes I caught myself having to look at the corner to distinguish between a seven and a nine.

The stratagem of arranging near identical marks in a pattern is an attempt to enlist shape in the service of number : if you recognize the shape you don’t need to count the dots. Shape recognition is distinction by type which the principle of distinction by number must displace in the cultural development of the species. For all that , even today, we feel at ease with shape and respond to it ‘naturally’ (perhaps because of the sexual instinct and childhood memories) while number is at first unappealing, it appears cold and  inhuman. The supposedly artificial distinction between the arts and the sciences is rooted in this primeval struggle between distinction by shape and distinction by number, a struggle which the latter is obviously  winning. In the pre-industrial past it was quite the reverse  : most ‘primitive’ tribes considered that distinctions of shape and thickness were so much more significant than numerical distinctions that they developed a large  and sophisticiated vocabulary to deal with the former while contenting themselves with half a dozen number words. And even in the domain of number itself shape cast its shadow: many societies used different number words  depending on the overall shape of the objects being counted. The Nootkans, for example, used special terms for counting round objects and traces of this practice persist in the ‘numerical classifiers’ of modern Japanese and Chinese¹.

A drawback of numerical distinction by patterning is that every new number requires its own special arrangement which must then be committed to memory. This certainly limits the range but the same patterns could be re-used with slight differences or could be combined in various ways. One would not have thought the effort involved was that great, not that much more than is involved in learning the alphabet. Also, the idea of familiarising people with numerals by way of card and board games is delightful (though presumably not done  deliberately). For some reason this promising system never got extended beyond six (otherwise we would have in our heads patterns for higher numbers) and taken in itself constitutes something of a dead-end in the history of numbering.

Domino numbers are a curious and attractive relic of days long gone. 

Q1. If we use dominoes as numbers, what base are they in? And what is the largest amount that can be represented by a single set of dominoes ?  

(To be continued)

“He who examines things in their growth and first origins will obtain the clearest view of them” (Aristotle).

The first thing to realize about bases is that Nature does not bother with them. Nature does not group objects into tens, hundreds, thousands and so on, not even into twos, fours, eights. Bases in the mathematical sense are entirely a matter of human convenience — Nature only uses base one. We are so used to thinking decimally and writing numbers in columns that we tend to consider that an amount, expressed in our modern Hindu-Arabic  positional numerals, is somehow truer than if it were expressed in, say, Chinese Stick Numbers. Even, there is a tendency to think that the modern representation of a quantity  is somehow numerically truer or more real  than the actual quantity  — an example of the delusionary thinking that doing mathematics all too readily gives rise to. Asked how many stones there are in a certain pile, an unhelpful but perfectly correct answer would be to transfer all the stones into a wheelbarrow and empty them out at the enquirer’s feet.
          So why do we bother with bases? Partly for reasons of space : an amount expressed in base one requires a lot of paper or wood or whatever material we are employing and this was and is an important consideration. But the main reason is that our perception of number is so defective that we find it very hard indeed  to distinguish between amounts of similar objects or marks beyond a certain very small quantity (seven at most). So what do we do? What we need is a second fixed quantity, a second ‘unit’, in terms of which we can assess larger quantities. What do we choose?
‘Ones’ are given us by Nature and more often than not this is all we are given. The sort of objects we shall want to assess either are collections of ‘ones’, like trees or sheep, or can be treated as if they were, e.g. mountains, villages and so on. We speak of ‘whole’ numbers thus implying that they are in some sense ‘entire’, ‘indivisible’. If our standard ‘one-object’ is a pebble it really is indivisible in a pre-industrial society and a cowrie shell, though it can be broken in two, ceases to be a proper shell if this is done. There is, in concrete number systems,  no question of dividing a ‘one’ and ‘fractions’ (‘broken numbers’), if defined at all, are represented by smaller objects or marks, not by splitting up the ‘one-object’. Whole number arithmetic was by Greek times firmly associated with the physical theory of atomism as put forward by Democritus who himself wrote works on arithmetic now lost.
A mass of ‘ones’ is perceptually unmanageable unless related to certain standard amounts we are familiar with. But Nature is extremely unhelpful in providing us with exact standard amounts. Litters of kittens and puppies are by no means standard, and apart from a slight prejudice in favour of the quantity five, the amount of petals in a flower or branches on a tree varies amazingly. Familiar fixed amounts such as the ‘number’ of hills on the skyline could only be of strictly local relevance and at the end of the day about the only available standard amounts given to man by Nature are the fingers which, counting the thumbs, come in fives. It is thus no surprise that number systems are dominated by the amounts five, ten and twenty.

The Secondary Unit

 What there is no doubt we do need and have done from very remote times is a secondary unit. This must be clearly distinguished from a base  since the latter is an extendable sequence of ‘unit’ sizes, a ‘geometric series’ like 1, 10, 100, 1000 and so on. Why didn’t ‘early’ societies (with some exceptions) go straight for the base system? The answer is that they didn’t need it and that it doesn’t come naturally, at any rate to practical people. For most purposes two ‘significant amounts’ or the first and second powers of the base are quite sufficient. Even one will do if it is of reasonable size because you don’t actually have to stop at the ‘square’ of the base as if it were a brick wall cutting off all access to a numerical  beyond. Fifteen hundred, apart from being more succinct,  sounds a good deal more natural than the pedantic ‘one thousand and five hundred’ which is what we ought to say by rights (Note 1). Without even defining the hundred  we could still cope perfectly well with quantities up to 999 reckoning in so many tens e.g. by speaking of 810 as eighty-one tens and a five. Generally we do not need to go anything like so far and the language is littered with sets of number words which, though they show base potential, peter out into nothingness without ever even making it to the ‘cube’ stage.
The ‘natural’ way of cultivating the wilderness is to clear an area and then when you’ve planted that, to clear another. Natural at any rate if you have to do a fair amount of the work yourself. If you are a conqueror you may, of course, have an eye on infinity and eternity from the start but this is  folie de grandeur. The first man to introduce wholesale decimalisation was the Ch’in Emperor and he is thought to have hastened his death by imbibing elixirs of immortality.
Numbers were measures before they became numbers ¾ even ‘one’ itself, the ‘father (better mother) of all numbers’, is essentially a measure, one drop, one mouthful, one foot. Our standard weights and measures are really only numbers that remain tied to particular contexts and functions. The Imperial system of liquid measures is an application of base two with four left out since the numbers involved are 1, 2 and 8.

                   1 pint     =  1 pint

                   2 pints   =   1 quart

                   4 quarts =  1 gallon  =  8 pints

Verbally, the number system stops here : although there are obviously larger quantities than the gallon we have  no special words for them, it is someone else’s job to work them out.

In our number words and pre-decimal measures we find a surface  order with an underlying picturesque confusion where all sorts of sets of numbers leave their traces. In the Avoirdupois weights we seem to have two sets of numbers, one proceeding from the ‘small’ end and one from the ‘large’ end, most likely developed by persons performing different functions. It makes sense to have the pound finely divided for sales over the counter to individuals, thus the appearance of the  large, but not too large, secondary unit 16 in sixteen ounces to a pound. From the wholesaler’s point of view we want a large quantity defined straight off, the hundredweight (which is not a hundredweight). We now quarter the hundredweight as it is always useful to divide something into four equal parts and we nearly but not quite converge with the rising 16 system. But there are not 32 pounds to a quarter but the anomalous 28. Is a systematic base system preferable even supposing we had a more suitable base than 10? Not necessarily for the people doing the work. Their principal concern is not logical  consistency but the ready availability of convenient set amounts which the chosen number system or systems should favour and promote. Moreover, once they have what they want, various landmark fixed amounts, they leave the system to its own devices.
There is certainly, within the context of a pre-industrial economy, no need for a number sequence stretching out into infinity :  on the contrary this very feature would have in many cultures provoked a certain malaise as indeed it still does to persons like myself. The idea of really large magnitudes is frightening like the idea of really large intervals of time. Although it is now unfashionable, even politically incorrect, to speak of cultures having specific traits, it is surely no accident that it was the Hindu mathematicians who gave us the first fully positional indefinitely extendable written number system. Indian thinkers, both Hindu and Buddhist, were obsessed with large numbers and vast spans of time : the kalpa for example is a period which lasts 4320 million years. Armed with the decimal base number system the Indians built ‘number towers’ reaching unimaginable heights and not only could they write down these quantities but they had names for many of them. In one legend the Buddha, challenged to list the numbers (read ‘powers’) beyond 10⁷, answers with the names for all the powers up to the colossal tallaksana  or 10⁵³ ¾ i.e. 1 followed by fifty-three zeroes (Note 2) . The Indian approach to large numbers is quite different from that of Archimedes who is, surprisingly in some ways, much more in line with the earlier ‘clear an area’ approach. He wrote a treatise on large numbers but showed none of the delectation and religious awe that the Hindu and Buddhist mathematicians clearly felt. Archimedes was concerned to show that the finite Greek number system could be extended upwards and outwards to deal with colossal quantities like the amount of grains of sand in the, for him finite, universe. But his aim was to tame the beyond not to lose himself in admiration of it. To the amazement of modern commentators, he did not quite hit upon the artifice of full positional notation.
The acceptance (or imposition) of a single indefinitely extendable base system has taken a very long time and is of comparatively recent date. For centuries individual and local numbering systems co-existed with the State imposed one, Roman or Napoleonic, especially in country areas. Until very recently by far the greater part of the inhabitants of Europe were illiterate and many of them used their own numbering systems and ‘peasant numerals’ like the notched sticks of Swiss cowmen. Inns could scarcely have carried on at all without the one-base slate and chalk system where the reckoning was totted up at the end of the evening or month, and in Spain it was once common for the innkeeper to toss a pebble into the hood of a traveller’s cape for each drink consumed. Today rustic numbering systems like the ‘milk sticks’ of cowmen in the upper Alps or the use of knots or pebbles  are things of the past : everyone has at long last agreed to at least write numbers in the approved standard decimal fashion. But for all that we do not think or feel in the way we write. In effect we still use the Babylonian secondary unit, sixty, for the subdivision of the hour, although we  express the quantity in a ten-base. We think ‘sixty minutes’ as a ‘chunk of time’  divisible into so many units, not as six tens of time. We do indeed have the availability of intermediate amounts, five minutes, ten minutes, quarter of an hour, but they are subordinate to the hour and the minute. A day is experienced as a unity which is in the first place decomposable into the unequal ‘halves’ of daytime and nighttime. We do not experience or think the day as two tens of time plus four units.

To the administrator, of course, the use of a consistent base-system is as necessary as the use of a ‘universal’ official language (Latin, English in the Commonwealth &c.). To him numbers have finally ceased to be tied to objects or activities, have become contextless,  in much the same way as, at a further level of abstraction, functions, to the contemporary mathematician, have ceased to be tied to numbers.

Alternatively, of course, we could start from the other end and opt for a ‘man-made’ secondary unit whose size would depend on our perceptual needs and what exactly it is we want to assess or measure. These three criteria 1.) availability of a standard amount; 2.) human perceptual  limitations and 3.) appropriateness for assessment purposes, conflict and one of the main problems of early numbering was how to reach an acceptable compromise between them.
Two is the first possibility for a ‘secondary unit’ but, although it has come into its  own in the computer era (because of the two states On and Off), it is clearly too small to be of much use for ordinary  purposes. A language spoken in the Torres Straits had a word for our ‘one’, namely urapan and a word for our ‘two’ okasa and that was about it. Their numerals went

         1.      urapan                  4.      okasa okasa
          2.      okasa                     5.      okasa okasa urapan
          3.      okasa urapan                

Understandably, since even a number as small as 11 would require six words, the natives referred to anything above 6 as ras — ‘a lot’ (Conant, p. 105).

A few things, or rather events, are viewed in threes witness phrases like “third time lucky” and we group quite a lot of things in fours (seasons, points of the compass &c.) but  the obvious first choice for a ‘secondary unit’ is five. Beyond five we really feel the need for a ‘secondary unit’ since  collections like  ½½½½½½ and ½½½½½½½ are  practically indistinguishable. Also, as it happens we have the five fingers to be able to check (by pairing off) whether we are separating out the items into groups correctly.

The Old Man of the Sea in Homer ‘fives’ his seals but for most herdsmen five would still have been rather too small as a secondary unit. So where do we go next? If we remain guided by the fingers the next possibilities are ‘both hands’ and what many primitive languages referred to as ‘the whole man’‘ i.e. ten and twenty.  Twenty is in some ways a better choice since, if we keep the option open of reverting to five for trifling amounts we can cope with very sizeable collections using batches of twenty. The Yoruba used twenty cowrie shells as their principal counting amount after the unit. Some modern European  languages which have long since become decimal show traces of an earlier vigesimal (twenty-based) system which probably suited farmers better. Hence Biblical terms like ‘three score years and ten’ in English and the French soixante dix-huit (sixty-eighteen).

A secondary  unit is, unlike the unit, not actually indivisible — since it is still made up of standard ones — so how do we keep it together if we are using objects as numbers? This depends on the choice of standard object and in practice is one of the motivations for the choice of object in the first place (or second place at least). Heaps of pebbles are heavy enough not to blow away but can all too easily be disturbed by people bumping into them, while piles of flat objects unless they are paper thin readily tip over and in any case really flat objects are hard to come by in nature. This is where shells are advantageous since if of the cowrie variety they stack up neatly and, even better, can be pierced and threaded on strings to make number rosaries. Beads make good numbers but since they are manufactured items they would not have been amongst the very earliest examples of object numbers.

However, on reflection I decide that introducing such a system would be premature. Within the bounds of a self-sufficient fishing, hunting or agricultural economy there would neither be any need for an indefinitely extendable number system nor would it have any special appeal. In the first place an inhabitant of such a society would not anticipate needing to assess really big quantities. Although a peasant needs more numbers than a hunter, in the past he probably rarely if ever needed numbers extending beyond about 400 , supposedly the upper numerical limit of a typical 19^th century Russian peasant. Large amounts of fruit, potatoes and so on would, of course, not be counted but be assessed by weight just like coins that we hand in to the bank. (Banknotes are still counted but in most banks the work is now done by a machine.)

The practical man, craftsmen, herdsman or farmer does not deal in ‘numbers’, he deals in fixed amounts that are significant in terms of his or her  daily work and/or perceptual apparatus. And such quantities do not usually  correspond to the transition points of an extendable base system like our own. To judge by the traces they have left on our language the two most popular ‘significant amounts’ beyond the unit in English speaking countries were, and to a certain extent still are, the dozen and the score. Although twelve as such is beyond our perceptual capacity, the image of two boxes each containing half a dozen eggs must by now have penetrated to the collective  unconscious, at any rate the English speaking one. Twelve, like ten, seems about the right size for making bundles or piles, but is better than ten in many ways because it can be halved, quartered and chopped into three equal portions. For dealing with larger amounts, the score which was originally a ‘score’ or notch a farmer made on a piece of wood as twenty animals passed through a gate, is about all you need so long as you know at least  twenty number words off by heart. The publisher of the red book travel guides to Europe is said to have counted the steps leading up to Milan Cathedral by transferring a pebble from one pocket to the other each time he mounted twenty steps.

A brief list of ‘significant quantities’ on a world-wide scale with reasons for their significance would perhaps be:

5                         hand, fingers,  right size perceptually

6                         half of dozen

10                        both hands, right size for base

12                        right size for base, many factors

 20                        hands and feet, multiple of 5 and 10

60                        many factors, multiple of all previous

100                    square of 10, many factors

Some of the above numbers are significant perceptually, notably 5 since this is around the stage when we cease to be able to assess objects numerically without counting them individually. Thus 5 combines significance because of its use in one-one correspondences (by way of the fingers) with significance as a perceptual ‘unit’. But it has the serious drawback that it cannot be divided up at all (has no factors). It is thus significant but not convenient as a ‘secondary amount’.
Having many factors is really more a matter of convenience than  significance as such but since previous significant amounts are amongst the factors of 60 and 100 these numbers acquire significance acquire significance by proxy. 60 is particularly rich in factors  and has the remarkable property of being a multiple of all previous significant amounts.
100 means nothing to us perceptually though it undoubtedly did to a Roman centurion who would have had in his mind’s eye the terrain covered by his infantry when lined up ready to give battle. 100 is around the ‘acquaintance’ mark, i.e. near the maximum number of persons one is able to relate to personally ¾ I believe I have read somewhere (Desmond Morris?) that 128 is about the limit and that this generally corresponds to the maximum number of persons one has in one’s address book.
But of course 60 and 100 are above all significant because of their divisibility ¾ the main use of 100 is in percentages though it still has the defect that one cannot divide it into three properly.  It must be stressed that a number’s divisibility is not just a matter of interest to modern number theorists : wholesalers or state suppliers receive commodities in bulk which they must subsequently sell or distribute to individuals and it is important that standard quantities should be easy to divide up. This is the reason why so many of the old Imperial measures are built around 20 or the powers of 2 ¾ as it is one of the main reasons why there is such hostility to metric weights and measures.
   One of the troubles with the transition points of a base-system, the ‘powers’ of the base, is that they are significant and convenient not in a practical but purely mathematical sense. Technically speaking, the unit, the base and its powers are the successive terms of a geometric sequence  1, b, b², b³, b⁴ ……. with common ratio b. My choice of twelve for the bundles that are to go into the second alleyway means setting b at 12. We would thus have 1, 12, 144, 1728… (since 144 = 12², 1728 = 12³). Now 144 and 1728 apart from being too large are not meaningful amounts in our day to day experience. The same goes for smaller choices of base.  5 is perhaps the most ‘significant amount’ of all in real-life terms but 5² = 25 is nothing special and 5³ = 5 ´ 5 ´ 5 = 125  even less. 6 certainly has some valid claims on our attention as a significant quantity but 36 has none.
One suspects that the success of a hundred as a ‘significant amount’ is due to its being a multiple of the significant amounts 5 and 20 ¾ it is actually 5 ´ 20 ¾ rather than it being the square of ten. A thousand is just a word meaning ‘large quantity’ and the ambiguous meaning of billion (a million millions or a thousand millions?) shows how vague such large quantities are. A million only has meaning with respect to wealth — and even this sense has been eroded by devaluation so that we find it necessary to replace the word millionaire by multi-millionaire which is even vaguer.

 Non-base extendable systems

Most people assume that once you have defined your ‘secondary unit’ you are somehow obliged to turn it into a true base (and I tended to think along these lines myself before writing this book). But of course you aren’t. The ‘tertiary unit’ or next standard amount we choose to define can be anything at all in principle. One of the few numerically advanced peoples that still used object numbers, the Yoruba,  took a pile of twenty cowrie shells as their ‘secondary unit’. They then combined five such piles of cowrie shells to make 100 in our reckoning, and combined two such piles to define their second most important amount after the unit, 200 in modern numbers. If they had operated a true base system the next halting point would have been 20² or 400 which presumably they considered too large. (Algebraically the Yoruba sequence goes 1, b, 10b and not 1, b, b² ). For a somewhat different, but nonetheless pragmatic reason, the Mayans, who also took 20 as their ‘secondary unit’, then moved on to 360 (instead of 400) for the next transition point in order to get close to the number of days in a year — or so it has been conjectured.

It does not in practice matter too much for addition and subtraction if the transition points are not in proper sequence (‘proper’ as we see it today) though it is desirable that they should be multiples of the first ‘significant amount’. Thus, anticipating a further fixed amount in my stick system I have already decided to opt for 60 since it is a multiple of 12 while 20 or 100 are not. Odd though it sounds, there is much to be said for defining a large ‘secondary unit’ and then defining  ‘units’ rather smaller instead of larger than it. This is in effect what the Babylonians did by taking 60 as principal amount after the unit which they noted as    . Such an enormous secondary unit makes it absolutely essential to have one or more halting points, or sub-bases,  in the intermediary territory which the Babylonians provided at the five and ten points. They defined the five transition by grouping the one-symbols and introduced a special mark for ten      but otherwise they used only the ‘one-symbol’ right up to 60 itself. For quantities > 60  the Babylonians proceeded by using 60 as a  true base, i.e. the next halt was at 60² = 3,600. In their case they seemed to have no misgivings about using such a huge  amount as tertiary unit which seems to contradict what I said earlier. But the Babylonian scribes who developed and used the sexagesimal number system were not hunters or herdsmen but officials helping to run a vast empire. They needed large numbers and spent their lives dealing in them as did the Egyptian scribes.

Practically speaking we require very different fixed standard amounts depending on the context. To divide a pound weight into sixty ounces would appear slightly crazy but we find it most convenient to divide up a fairly short interval of time, the hour, into sixty minutes, while we divide up a somewhat larger interval, the day, much less finely. The numbers 60, 12 and 24 are not imposed on us in the way the number of days in the year is : we could divide up the day into 60 or 72  or any (even) number of ‘hours’ and divide each ‘hour’ into 12 or for that matter 17 ‘minutes’. The unsystematic way in which we divide up the day seems right : there is, as far as I know, no SI project to decimalise time (though the ancient Egyptians did just this) and the very idea fills me with horror.

___________________________

Note 1

“Yet it was the Indians who reckoned the age of the Earth as 4.3 billion years, when even in the 19th century many scientists were convinced it was at most 100,000 years old ( the current estimate being 4.6 billion).” The apparent source for this is: Pingree, David, Astronomy in India, in Astronomy Before the Telescope, p.123-42.  Quoted Chasing the Sun by Richard Cohen, p. 132

¹  I recently came across an interesting example of how restricting the idea of always keeping to a base is. I noticed, or had read somewhere, that the Binomial Coefficients were powers of 11 and this made sense since they can be defined by starting with 1 and getting the next term by shifting what you’ve got across one column and adding. Thus

          1                             1  =   11⁰

                                      1        0 

                                      1        1                            11  =  11¹

                             1        1        0  

                             1        2        1                           121  =  11²

                   1        2        1        0

                   1        3        3        1                         1331  =  11³

          1        3        3        1        0

          1        4        6        4        1                       14641  =  11⁴

 However, what happens now?  The next line of Pascal’s Triangle is supposed to be

                   1        5        10      10      5        1 

 This isn’t a power of 11 surely.  But who says you can’t overstep the base if you want to?

 (1 x 10⁵) + (5 x 10⁴) +  (10  x 10³) +  (10 x 10²) +  (5 x 10) +  1 

  =   100,000 + 50,000 + 10,000 + 1,000 + 51   = 161,051   = 11⁵

Classifiers      In other cultures different bases were used depending on the different objects being counted. Flat objects like cloths were counted by the Aztecs in twenties, while round objects like oranges were counted in tens.  The use of classifiers obviously marks an intermediary stage between the era when numbers were completely tied to objects and the era when they became contextless as now. We still retain words like ‘twin’ and ‘duet’ to emphasize special cases of ‘twoness’, note also ‘sextet’, ‘octet’ &c.     The complete dissociation of verbal and written numerals from shape and substance is today universally seen as ‘a good thing’ especially by mathematicians. But classifiers were doubtless once extremely useful because they emphasized what people at the time felt to be important about certain everyday objects and activities, and they remain both a picturesque reminder of the origins of mathematics in the world of objects and our sense-perceptions. The removal of all such features from mathematics proper seems to be a necessary evil but at least let us recognize that it is in part an evil : the banning of contextual meaning from mathematics, the language of science and administration, is typical of the depoeticization of the modern world.

Base sixty
It is not known why the Babylonians chose 60 as their most significant amount after the unit. The fact that 60² = 360  is close to the number of days in the year may have something to do with it. Certainly, 60 would not have been the original choice. It has been suggested that 60 evolved as a compromise solution to the separate claims of 5 and 12  which were already well established as ‘secondary units’ within the territories conquered by the Babylonians. Since 60 has as factors all the main bases and significant amounts smaller than it, including 10 and 20, it had something for everyone : it was a numerical Pax Romana.