“I should have liked to produce a good book. It has not turned out that way”

Wittgenstein wrote in the Preface to the Investigations that he would have liked to write a good book, but it didn ’ t turn out that way. This may superficially seem to be false modesty, given that what he wrote is a masterpiece. This paper argues that it is not false modesty, and attempts to pin down various flaws in the book, some structural and others not. These include the opening quotation from Augustine, the thin character of language game 2, the rule following considerations, the private language arguments, and the poorly located, well-disguised and over-compressed discussion of the pictoriality of the proposition and critique of the picture theory of meaning in §§428 – 65.


The Preface
The final paragraph of the Preface to Wittgenstein's Philosophical Investigations runs as follows: "I should have liked to produce a good book. It has not turned out that way, but the time is past in which I could improve it." At first blush, this may seem undue modesty, given that the work has been widely viewed as a masterpiece. Some explanation of his reasons for this judgement are mooted in the Preface. In the first paragraph he remarks: "it seemed to me essential that in the book the thoughts should proceed from one subject to another in a natural, smooth sequence." In the next paragraph he continues: "After several unsuccessful attempts to weld my results together into such a whole, I realized that I should never succeed." The reason was that the best he could write would never be more than philosophical remarks, for as soon as he tried to force his thoughts into sequential argument, they grew feeble. So what he produced was akin to sketches of a landscape, the better of which he has selected for inclusion. So the book is in fact only an album.
In earlier drafts of the Preface, Wittgenstein made his qualms clearer. On 15 th September 1937, he wrote that when he "thinks for himself" as it were, without the idea of writing a book in mind, he circles around a topic. That is his natural form of thought and being forced to think sequentially is torture. He is wasting untold labour on an arrangement of his thoughts which may be worthless. The next day, he wrote a short Foreword in which he explains that he was unable to thread his thoughts into one string of consecutive argument. His attempts to do so had failed, and the only presentation he could give was to connect his remarks by a network of numbers that would make clear their extremely complicated connections. "May this be taken instead of something better -which I should have liked to produce." It is striking that this "network of numbers" that would make clear "their extremely complicated connections" is reminiscent of Wittgenstein's predicament over the composition of the Tractatus and his invention of the very complicated structure of that book, devised in MS 104 in the form of a logical tree or hypertext. 1 In the Tractatus the first levels of the tree-structure were written de novo. In MS 104, the first 280 propositions were not transcribed from Wittgenstein's notebooks, but were composed on the hoof, as it were, to create a logical tree onto which all his satisfactory notebook material could be assigned a place signified by a complex location-indicative number. The result is undeniably a work of art with a beautiful structure. It seems clear that Wittgenstein harboured no comparable intentions with respect to the first draft of the Investigations. It is not, however, at all clear what he did have in mind.
In Norway in late 1936, Wittgenstein began the composition of a first draft (MS 142: the Urfassung) of the Investigations. It corresponds roughly to § §1-189(a) of the Investigations. It is written as consecutive prose without any enumeration. This was typed up in May or December 1937 (now known as TS 220). The enumeration was inserted later, as is evident from pages 78-91. These were deleted and redrafted on pp. 91ff. Pages 78-91 lack any enumeration, so the numbering must have been inserted after the deletion of these pages and the composition of pp. 91ff. In January, in MS 117, Wittgenstein turned to the continuation. This was the first draft of what became TS 221, a version of Part 1 of the Remarks on the Foundations of Mathematics. TSS 220 and 221 conjunctively are now known as the Frühfassung or Early Draft of the Investigations. Note that there was smooth transition from the former to the latter. Wittgenstein's original idea was to use the complex considerations of MS 142 on meaning and use, ostensive definition and explanation, vagueness and family resemblance, philosophy, rule-following, reading and understanding as the foundation for an investigation into the nature of logical and mathematical necessity and the demolition of the pretensions of metaphysics.
All this is of importance, both for the understanding of the Philosophical Investigations and for Wittgenstein's persistent dissatisfaction with what he was producing and, later, had produced. The 1937 draft of a brief foreword further confirms that he had imposed no enumeration on TS 220 (MS 142), but initially presented his remarks as consecutive prose, with which he remained unhappy. What he proposed now was to continue with remarks on mathematics and logical necessity that would be accompanied by a complex system of enumeration. He wrote as follows: I start this book with the fragment of my latest attempt to arrange my philosophical thoughts sequentially. This fragment has, perhaps, the virtue of making it comparatively easy to obtain a grasp of my method. I'll continue this fragment with a mass of more or less loosely ordered remarks. But the links between these, where the arrangement does not make them evident, I shall clarify by means of a numbering system. Each remark will have a serial number, and apart from that, the numbers of those remarks which are related to it in important ways. (TS 225,II) Whether Wittgenstein ever attempted to construct such a prima facie clumsy and cumbersome system of enumeration is unknown. If he did, it would have been made on separate sheets of paper that he would have destroyed when he realized that the idea was flawed.
So, one fundamental source of unease with the Investigations was absence of consecutive argument -an album of sketches rather than a finished canvas. A consequence of this is that many points may have a first sketch done from one perspective at a given point and a later sketch from a different perspective fifty or a hundred remarks later. This creates formidable difficulty in giving an overview of his ideas on many topics.
Another source of unease was the adequacy or inadequacy of his selection of remarks from his typescripts. As he wrote in TS 225, "I wish that all these remarks were better than they are. They are lacking -to put it briefly -in power and precision. I publish those here which do not seem to me too barren" (TS 225, III). As we shall see, his attempts at selection do not always seem wise and he sometimes excluded wonderful observations. This is especially true of his remarks on philosophy in the Big Typescript.
A third flaw is not mentioned but is a corollary of his idiosyncratic method of composition. The rhythm of the text is altogether erratic. In some places the argument proceeds slowly and steadily -an obvious example is the discussion of reading in § §156 -178, derived directly from the Brown Book and its German redraft and continuation: Eine Philosophische Betrachtung. By contrast, elsewhere the progress is sometimes positively frenetic, as Wittgenstein rushes down one side street after another -an example is § §243-315 which constitute the "private language arguments". 2 Just how erratic this becomes is made evident in the tree diagrams in my exegesis of Investigations § §243-427 (in Hacker 2019).
In this paper I shall attempt to identify flaws, especially, but not only, structural flaws in the Philosophical Investigations. The book remains, I believe, one of the greatest masterpieces in the history of philosophy.
But the flaws are non-trivial. Bringing them into the light does, I hope, facilitate understanding. They are, to a large extent, the price Wittgenstein paid for his idiosyncratic method of creation. Perhaps the natural form in which his genius expressed itself was ill suited to the general form of philosophical books, and his inability -once he had thought through a topic -to take a clean sheet of paper and start writing it down in consecutive prose was the price he paid for his genius.

2.Beginning the book
Wittgenstein opened the book with a long Latin quotation from St Augustine's Confessions I. 8. It runs as follows in translation: When grown-ups named some object and at the same time turned towards it, I perceived this, and I grasped that the thing was signified by the sound they uttered, since they meant to point it out. This, however, I gathered from their gestures, the natural language of all peoples, the language that by means of facial expressions and the play of eyes, of the movement of the limbs and tone of voice, indicates the affections of the soul when it desires, or clings to, or recoils from, something. In this way, little by little, I learnt to understand what things the words, which I heard uttered in their respective places in various sentences, signified. And once I got my tongue around these signs, I used them to express my wishes. 3 Wittgenstein uses this quotation to exemplify a referential conception of language, which in one form or another, has dogged philosophy since its inception. It is also a natural way to think about words and sentences, an erroneous way of thinking to which we are intuitively inclined. Why did he select this particular passage? He explained that this conception is significant for us precisely because it belongs to a naturally clear thinking person temporally far removed from us, who does not belong to our cultural milieu (MS 111,15). Wittgenstein explained that he decided to begin with the quotation from Augustine not because he could not find this conception as well expressed by others, but because the conception must be important if so great a mind held it (Malcolm 1984, 59f.).
Nevertheless, the quotation is problematic because too much of it accords with Wittgenstein's own conception of language learning and acquisition.
(i) when adults name some object towards which they turn, the object meant is indeed signified by the uttered name. (ii) gestures are indeed the natural language of mankind. (iii) facial expressions, play of eyes, movement of limbs and tone of voice do indeed indicate human affections (although not, as Augustine puts it, affections of the soul). (iv) the child is indeed exposed to words uttered in their respective places in various sentences. With this Wittgenstein can have no quarrel, and these points cannot be the reason for selecting this quotation as representing a natural and misguided way of conceiving of the nature of language and the meaning of words. To this extent, the quotation from St Augustine is misleading for the reader. It is meant to incorporate a picture of the essence of language, to represent the idea that the meaning of words is seen as the foundation of language (MS 152,40), and to make it appear as if naming is the foundation and essence of language (BT 25). But it is difficult to apprehend this picture in the passage Wittgenstein quotes, although, to be sure, Wittgenstein goes on to explain what he finds objectionable by extracting a picture and a conception of language.
From Augustine's description of how he must have learnt to speak (based he said, on his own observation of children's early uses of language), Wittgenstein himself extracts a pair of apparent truisms 1. Words name objects 2. Sentences are combinations of names This may be called "Augustine's picture of language". It is held to be intuitive and pre-theoretical. From these apparent truisms, Wittgenstein himself derives a more sophisticated idea or conception that consists of three further contentions: 3. Every word has a meaning 4. A word is correlated with its meaning 5. The meaning of a word is the object it stands for.
This may be called "the Augustinian conception of language". (1) -(5) provide the main themes of Investigations § §1-59. However, Wittgenstein and his readers have to work hard to see the Augustinian conception of language in the quotation, since it is inextricably interwoven with ideas that are too close to Wittgenstein's own. To be sure, many other thinkers emphasized the primacy of names and naming, and the connection of names in a sentence in which a named attribute is predicated of a named subject. So, for example, Hobbes: But the most noble and profitable invention of all other, was that of SPEECH, consisting of names or appellations, and their connexion; whereby men register their thoughts; recall them when they are past; and also declare them to another for mutual utility and conversation 4 or John Stuart Mill: It seems proper to consider a word as the name of that which we intend to be understood by it when we use it 5 or John Locke: [children] begin by degrees to learn the use of signs. And when they have got the skill to apply the organs of speech to the framing of articulate sounds, they begin to make use of words . . . The verbal signs they sometimes borrow from others, and sometimes make themselves, as one may observe among the new and unusual names children often give to things in the first use of language. 6

Language-game 2
In §2 Wittgenstein introduces the idea of a language for which the description given by Augustine is right.
The language is meant to serve for communication between a builder A and an assistant B. A is building with building stones: there are blocks, pillars, slabs and beams. B has to pass him the stones and to do so in the order in which A needs them. For this purpose, they make use of a language consisting of the words "block", "pillar", "slab", "beam". A calls them out; B brings the stone which he has learnt to bring at such-and-such a call. --Conceive of this as a complete primitive language.
The difficulty lies in the final sentence. It is repeated in §6: "We could imagine that the language of §2 was the whole language of A and B, even the whole language of the tribe. The children are brought up to perform these actions, to use these words as they do so, and to react in this way to the words of others." The speech activity is characterized as a language-game in §7.
It is striking that Wittgenstein himself later raised the question of whether he had not made things too easy for himself in language-game 2 (MS 165, 95). I believe he had. First, one cannot conceive of a society that can cut blocks, pillars, slabs and beams and build unspecified buildings with them, but engages in no other social activities involving communication (going home after work and being greeted, eating together and asking for the food to be passed or complimenting the hostess on the food, growing food and harvesting it together, and so forth). Wittgenstein does not specify what purpose these buildings have: are they dwellings? In which case who gets them? Are there property laws or conventions without any words to specify them? Is that intelligible? Or are they covered markets? Or temples? Whatever they are, it seems inconceivable that they can obtain without a much richer language. Nor can one conceive of a society that has mastered the arts of building, but no other constructive arts: making clothing, cooking, making tools and utensils. Wittgenstein insists that to imagine a language is to imagine a form of life ( §19), but he has not described the form of life of the speakers of language-game 2.
Secondly, it is inconceivable that there be a language consisting only of four names of building stones. Granted that this is what children are taught, is there no sign of negation for purposes of correction and denial? No doubt the concept of a language is fluid, but surely more is needed than this? Are there no gestures ("the natural languages of all people")? And do the gestures not have a meaning?
Thirdly, language-game 2 was supposed to be an example of a more primitive language than our languages, but one which Augustine's description fits. But little reflection is needed to realize that neither the Augustinian conception nor Augustine's picture are exemplified here.
For (i) there are no multi-word sentences, so it cannot be said that sentences consist of combinations of words. Indeed, in §19 the distinction between word and sentence is argued not to apply to language-game 2. (ii) Even more strikingly, the meanings of 'block', 'slab', 'pillar' and 'beam' are not the objects they stand for (PI §1(b), final sentence). If they were, it would make sense for us, as observing anthropologists, to say that the meaning of 'pillar' is standing on the meaning of 'slab', or that the meaning of 'slab' has just broken. But that makes no sense.
These are surely design flaws. But we should stand back and consider whether they matter greatly. What Wittgenstein is trying to do is to get his readers to focus upon an anthropological or ethnographic conception of language by contrast with the calculus conceptions of language that characterized the Tractatus, as well as Frege's and Russell's distinctive philosophies of logic and language. He is embarked upon a campaign to eradicate the notion that "In the beginning was the Word" and to replace it by the idea that "In the beginning was the Deed". Language should be viewed as a human activity, integrated into the hurly-burly of life. Words are deeds. They are used to do things in human interaction and discourse. The meaning of a word is what is given by an explanation of meaning; it is what is understood when one grasps the explanation; an explanation of word-meaning is a rule for the use of the word; the meaning of a word is its use (how it is to be used). The campaign is conducted on many fronts, but it is surely successful once understood, despite the flaws in the opening remarks of the book.

4.The "rule-following considerations"
In §143 Wittgenstein begins a new theme or set of interwoven themes, which dominates the next hundred sections. The initial setting of the problem is to reconcile the fact that one normally understands a word in use at a stroke, although the use of the word (its applications in discourse) is, so to speak, spread out in time in all its multiple applications. How can something that is spread out over time be understood in an instant ( §138)? In §143 Wittgenstein introduces the pivotal example with which he works -a case in which A gives B the order to write down a series of signs according to a certain formation rule, for example the order to expand the series of even integers, i.e. the order to write down the series "+2", i.e. the order to continue the series "2, 4, 6, 8, 10, 12, and so on". Many issues are interwoven in the sequel, for example, how does the order, or formula, or rule for the series determine the series? What is the nature of understanding the rule (of not understanding and of misunderstanding)? If B continues the series up to 1000 and then continues "1004, 1008, 1012, …", with what right can we say that he has misunderstood the order or rule? Can he not reply that he is doing the same after 1000 as he was doing up to 1000? Has he not interpreted the rule to mean "Add 2 up to 1000, and thereafter add 4"? So does every rule require an interpretation? Wittgenstein tries out a variety of misguided moves: 1. That B, in continuing "1004, 1008, …" is not doing the same, whereas going on "1002, 1004, …" is doing the same after "1000" as he was doing before "1000". 2. That he is doing what the teacher meant him to do. 3. That the steps are determined by the formula. 4. That each step in continuing the series requires an intuition. 5. That the rule of the series, which we grasp at a stroke, contains its applications. 6. Understanding a rule of a series is a mental state that contains what is understood: namely how to go on. 7. That going on requires an interpretation, so understanding a rule involves interpreting it. (1) invokes the concept of identity to resolve the puzzle. But this is futile (compare "It is five o'clock on the sun when it is the same time on the sun as it is when it is five o'clock here" (PI § §350-1)). (2) appeals to speaker's meaning as determining what is correct (anticipating a Gricean move). But this misconstrues meaning something as a mental act that anticipates all B's applications of the rule. (3) involves a misuse of the notion of determination in this context, since all it can legitimately do is differentiate rules that have a single correct application, like "+2", and rules that don't, like "2". (4) contains a peremptory repudiation of Brouwer's intuitionism (if you need an intuition of two-oneness to expand the series of natural numbers, then you also need one to expand the series "0, 0, 0, 0 …"). (5) is a move against mathematical Platonism according to which the solution to a mathematical question is already "there", contained in the rule, which we have to unfold. (6) is part of a persistent critique of the conception of understanding as a state as opposed to an ability. Abilities cannot be said to contain their exercise. (7) contains criticism of the idea that became popular after Wittgenstein's death, when Donald Davidson argued that all understanding is interpretation. All these moves are misconceived, misconstruing the nature of following rules.
Two strategic points should be noted: (a) the whole debate culminates in the insight that following a rule is a practice, an idea pivotal both for the private language arguments that follow §243 and for Wittgenstein's philosophy of mathematics in RFM Part 1. (b) the example of expanding the rule of a series is tailor made for the philosophy of mathematics, as is evident in RFM 1.
(a)For someone to follow a rule presupposes a regularity. There cannot be just one occasion in the history of mankind in which someone followed a rule. Following a rule is not an "one-off" affair. There must be a regular use of the rule as a standard of conduct, appeals to the rule as warranting what one did, as invalidating what someone else did. One may take as a persuasive example rules of orthography or such simple examples as "Don't say 'He were', say 'He was'". So, if a rule is being followed then there is a custom of regular behaviour. Further, there is an activity that exemplifies a regularity that is recognized as a uniformity. Moreover, that uniformity must be recognised as a norm, a standard of correctness. In short, rule-following is a normative practice. With us human beings, it is a social practice.
(b)the upshot of the long and elaborate investigation is no less pertinent to Wittgenstein's philosophy of mathematics for which it was crafted than it is to his philosophy of psychology and the private language arguments. But the working example of expanding an arithmetical series, while pivotal for the investigation of logical and mathematical necessity in RFM 1, is arguably inappropriate for the discussion of the meanings of words other than number words and of the essential logical publicity of natural language (PI § §243ff.) For this what is needed is an example of applying a humdrum rule of language, not a mathematical rule for the use of numbers and number words.
Evidently, Wittgenstein could not bring himself to redraft the rulefollowing considerations in 1943/4 when he decided not to continue the draft with RFM 1. What he would have had to do is take an example of a simple rule for the use of a word and put it through similar paces to yield the same conclusion, viz. that natural language is a rulegoverned practice in a human community. 7 So, one might take as an example an ostensive definition of a colour-word, as in "That → is magenta" or more explicitly "That → colour is magenta". This is not an empirical statement like "That cushion is magenta" or "My favourite colour is magenta" or "The dramatic sunset ranged from gold to magenta". It is a rule, an ostensive definition, of a colour word. We apply the rule when we predicate 'magenta' of a physical object, and we appeal to the rule when we say, for example, "I said the cushions are magenta because they are THAT → colour (pointing at the sample in a colour chart), or when we say, "I asked for six yards of magenta silk, but this silk is not magenta. Look, it is not THAT → colour." But now, if someone uses the word 'magenta' as we all do, but tomorrow starts using magenta in describing things that are Oxford Blue, can we gainsay him? He may explain that 'magenta' in his language, means magenta until tomorrow and from tomorrow onwards it means Oxford Blue. He may insist that he is using "magenta" in exactly the same way as before. It is of great interest that in 1936, Wittgenstein, en passant, came up with something very close to what he would have needed in order to transform the rule-following considerations from focusing on a rudimentary arithmetical rule to focusing on a rule for the use of a word in natural language. Imagine colour words used so: People have seven colour words and they have seven days in a week; and they use the words in a cycle, so that they use "blue" on Tuesdays for that which they used "red" on Monday, etc. Does one then use the word in a different sense each day? They mean the same. But they do not mean the same colour. "Tomorrow" means the same every day. But it does not mean the same date. So they go on with the word "blue" to a different colour every day. But does "blue" always mean "blue"? --yes, of course. What else could it be?
They might give an ostensive definition of "blue", and then they would go on automatically to call this other colour "blue" tomorrow. --They might on occasion ask, "but you said yesterday that was blue, how can it be today?" (LPE,311) This example of a rule of natural language might be used in much the same way as the rule for continuing a series to show that following a rule is a normative practice presupposing a regularity, recognised as a uniformity, and treated as a norm of correctness.
It is evident that a very similar critical discussion of the misguided moves that Wittgenstein introduced in his rule-following considerations could be constructed here too in the case of a linguistic rule. Let us suppose that the teacher A gives his pupil B an ostensive definition of red by pointing at an object and saying "That is red". Now suppose that B goes on to call differently coloured objects 'red' in a weekly cycle. What makes that wrong? In virtue of what can we say that the pupil has misunderstood his teacher and misapplied the rule? Here we can duplicate (1) -(7) above: 1. That B, in predicating red of differently coloured objects every day of the week is not doing the same with 'red' and that is why he is wrong. 2. That B's calling only red things 'red' would be doing what his teacher A meant him to do on each application. 3. That the rule "This is red" determines that going on to call each red object 'red' is correct. 4. That each application of 'red' to an object requires an intuition. 5. That the rule "That is red" contains all its indefinitely many applications (as a Fregean concept determines its extension, in as much as were its extension different, it would be a different concept). 6. Understanding the rule "That is red" is a mental state that contains the knowledge of how to go on in all future cases.
7. That every application of the rule "That is red" requires an interpretation. To which one might add 8. That what makes any given application of the rule "That is red" correct is that everyone else says that the given object is red: red is what most people call red. 9. Red is what seems to most people to be red. In this way the rule-following considerations might be reconstructed in a manner designed to be continued by the private language arguments rather than by the exploration of logical and mathematical necessity.

The Private Language Arguments
I have already observed that the private-language arguments ( § §243-315) are both erratic and frenetic, jumping from one topic to another at bewildering speed. There is, however, a more serious structural flaw here. The introduction of the problem in §243 is unnecessarily opaque. A "private" language is held to be one in which "the words refer to what only the speaker can know -to his immediate private sensations. So another person cannot understand the language." The question is raised as to whether such a language is conceivable. Is it intelligible that there be such a language? This introduction leaves it obscure why the question is of any interest. After all, just off the cuff, why should anyone care whether such a language is intelligible or logically possible. Our common-or-garden languages, English, German, French, are certainly not like that, for they are understood by millions of people. The pivotal point is that a very large number of philosophers throughout the ages, and many contemporary psychologists and neuroscientists, have inadvertently committed themselves to an array of assumptions that imply that the languages we severally speak are logically private. So a critique of the notion of a logically private language is at the same time a deep criticism of diverse philosophies of language and of psychological and neuroscientific presuppositions.
The introduction of the idea of a logically private language is unhappy. The grounds for the idea of logically private language are widely held misconceptions. Wittgenstein's introduction of these grounds is itself too compressed. We are very quickly introduced to the idea that one misguided sense in which sensations are private is that only I can know whether I am really in pain, another person can only surmise it ( §246). This may be called "epistemic privacy". Another misconceived sense of privacy of sensations is that another person can't have my pains. This may be called "private ownership" of pain. A third sense is that only I can give a private ostensive definition of 'pain' by mental pointing at what only I have and only I know.
These are the three legs upon which the illusion of a private language rests. The first two force the third upon us. Certainly, this could have been spelled out more clearly. Strikingly, very little attention is given to the first two senses, the main focus being on private ostensive definition of sensation words. In particular, private ownership of sensation is given very short shrift ( § §253-4). Indeed, such short shrift that many distinguished interpreters (e.g. Strawson, von Wright, Kenny) took it for granted that Wittgenstein meant that two people cannot have the numerically identical pain, but only qualitative identical pains. That is exactly the opposite of what Wittgenstein meant, which was rather that that very distinction, apt for public objects, has no application to sensations. Sameness of pain is, in this respect, similar to sameness of colour, weight, or height. One cannot say that the sofa and the armchair cannot have the same colour, since the colour of the sofa belongs to the sofa and the colour of the armchair belongs to the armchair. That would be to treat belonging to the chair as an identifying feature of the colour, as if the colour were a substance and belonging to the chair its specific difference. He spelled this out in the Blue Book: We use the phrase "two books can't have the same colour", but we could perfectly well say: "They can't have the same colour, because, after all, this book has its own colour, and the other book has its own colour too". This also would be stating a grammatical rule --a rule, incidentally, not in accordance with our ordinary usage. The reason why one should think of these two different usages at all is this: We compare the case of sense data with that of physical bodies, in which case we make a distinction between: "this is the same chair that I saw an hour ago" and "this is not the same chair but one exactly like the other". Here it makes sense to say, and it is an experiential proposition: "A and B couldn't have seen the same chair, for A was in London and B in Cambridge; they saw two chairs exactly alike".
(Here it will be useful if you consider the different criteria for what we call the "identity of these objects". How do we apply the statements: "This is the same day …", "This is the same word …", "This is the same occasion …", etc.?) (BBB, So too in the case of "He can't have my pain, because his is his and mine is mine" -they cannot be the identical pain but only similar: our pains are numerically distinct but qualitatively identical. But that very distinction does not apply to pains any more than it applies to heights, lengths or colours. It is at home with physical bodies, but not with pains. A further defect in the exposition in the Investigations is that the scope of the argument is left very unclear. Wittgenstein's working example is the sensation of pain, presumably because here the illusion is at its very strongest. Later it becomes clear that "inner experiences" are likewise included ( §256), as are colour impressions ( § §273-4). Elsewhere Wittgenstein runs through variants of his arguments in the Investigations and includes sense-data, experiences, sense impressions and objects of consciousness.
None of these faults, if faults they be, derogate from the brilliance of the private language arguments. But rectifying them would have made the understanding of the arguments very much easier and would greatly have reduced the extensive misunderstandings.

Investigations § §428
-65 § §428-65 contain some of the earliest drafted remarks in the whole book. Many of them predate the Big Typescript of 1933. The great themes with which they deal had occupied Wittgenstein's mind since the nineteen-tens and his first engagement with philosophy. The remarks in § §428-65 are mostly taken from his 1929-32 critical reflections on the fundamental ideas that informed the Tractatus' vision of language and logic, logic and reality, thought and language. They were reworked and transcribed more than most other parts of the Investigations, first in the notebooks that were sifted for the "Early Big Typescript", then for The Big Typescript, subsequently for the Umarbeitung, the reworking of The Big Typescript, then for Volume XII, MS 116. They were again selected for inclusion in the draft of the Investigations compiled in 1945, in which he added all the remarks from §421 to the end. So Wittgenstein had struggled with these ideas since 1929, thought them through indefinitely many times, copied them out indefinitely many times. They were, I suspect, so familiar to him that he altogether forgot what may reasonably be demanded of an intelligent reader working as hard as he can upon the text. Wittgenstein wrote in the Preface that he didn't want to save anyone the trouble of thinking for themselves, but in these remarks he arguably gave them too few clues to understand what he was up to. Moreover, the argument is excessively dense, different (though related) ideas follow on too swiftly for comprehension -they seem to tumble over each other (the contrast with the discussion of reading in § §156-78 could not be greater) -and it is exceedingly difficult to see what he is talking about and what his targets are. It is small wonder that for many decades these were the least understood parts of the book, as is patent from most of the reviews of the book by distinguished writers such as Strawson and Malcolm. In fact, as we shall see, these remarks contain the deepest criticisms of the Tractatus' metaphysical vision of the relation of thought, language and reality, and the most profound criticism of the picture theory of meaning. It should surely have been placed in the early parts of the book prior to §89. For most of the remarks in § §1-89 are concerned with criticising the Tractatus on words and names, names and naming, logically proper names, simple objects, ostensive definitions and explanations of meaning, meaning and use, vagueness and determinacy of sense. It would have been appropriate to locate § §428-65 here, and to explicitly identify the target as the picture theory of meaning and the Tractatus account of the intentionality or pictoriality of the proposition, namely how it is possible for a proposition to be false yet meaningful. It seems to me to be a dire flaw in the structure of the Investigations.
This group of remarks opens with the idea that when we come to think about it a thought is an extraordinarily strange thing. How can a thought anticipate reality, when what it anticipates may not yet exist, as when we expect A to come before he comes. How can it prefigure reality when what we think may never be the case, as when we think falsely that things are thus-and-so? But this was one of the pivotal problems of the Tractatus which the so-called picture theory of meaning was constructed to answer, namely: how can a proposition or thought be false but meaningful, how can it depict what is not the case? And how can one think of something that isn't here: of NN, who is in New York, or of Alexander the Great, who no longer exists, or of Father Christmas, who never existed? It seems as if thought is like a ballistic missile than can pinpoint things at a distance and things in the past and present, but also things in the future and things that do not exist. The solution to this in the Tractatus lay in the ontology of simple objects and in the doctrine of analysis, on the one hand, and in the harmony between language and reality, on the other. This harmony was conceived to obtain as a consequence of the isomorphism between a thought or proposition and the state of affairs, the possibility, that it depicts. That itself is a consequence of logically proper names naming simple, sempiternal objects in reality and of the doctrine that the logical form of the name corresponds to the metaphysical form of the object it stands for. Indeed, it is this harmony that is the subject of the next critical remark in Investigations §429. But it would have helped readers greatly to have been told this, rather than being left in the dark.
§429 opens with the assertion that the agreement, the harmony, between thought and reality consists in this, that if I say falsely that something is red, then all the same, it is red that it isn't. It adds the further explanation that if one wants to explain the word 'red' to someone, in the sentence "That is not red", I do so by pointing to something that is red. This wonderfully exemplifies Wittgenstein To understand what is at stake here, we must revert to the Tractatus. The picture theory of the proposition explained the conditions of the possibility of representation in terms of an agreement of form (a harmony) between a proposition (or thought) and what it depicts, irrespective of whether it is true or false. That isomorphism is ensured by the correspondence between the simple unanalysable names of which a fully analysed proposition consists and the simple sempiternal elements that constitute the substance of the world. The logicosyntactical forms of simple names mirror the logico-metaphysical forms of objects, and the elementary proposition is isomorphic with the possible fact the actuality of which would make it true. All this, and much else that flows from it, was repudiated in the early 1930s and is repudiated here and in the following remarks. A rule for the use of signs was misinterpreted as a connection between language and reality, namely -"the proposition that p" = "the proposition which is true if and only if p" It is true that one can read off a proposition that p the fact that makes it true, but that is not due to a metaphysical harmony between language and reality, but is merely a move in grammar licensed by a substitutionrule: It is true that we explain what 'red' means by pointing at something red, but far from this connecting language and reality, it is an ostensive definition of 'red' that incorporates a red object, used as a sample, into the means of representation. The sample, thus used, belongs on the side of what does the representing, not on the side of what is represented. So there is, in the requisite sense, no connection between language and reality (that does not mean that we do not refer to reality when we speak; it means that there is no pre-established harmony between language and the world). It would have helped readers of the Investigations to have been given a few hints of all this, and it was unreasonable to have expected them to see it unaided.
The next few remarks are concerned with the idea that signs by themselves are dead, mere noises or marks on paper. It is tempting to think that what gives signs their life are mental acts of meaning and intending, on the one hand, and of understanding, on the other. That was indeed what the author of the Tractatus had suggested. "Thinking the sense of the proposition" (TLP 3.263) is what projects the proposition onto reality. Hence Investigations §431: "There is a gap between an order and its execution. It has to be closed by a process of understanding." "Only in the process of understanding does the order mean that we are to do THIS. The order --why, that is nothing but sounds, ink-marks. -" Similarly, according to the Tractatus, only in the process of meaning does a sentential sign describe the state of affairs the actuality of which make it true. § §433, 437, express puzzlement that an order and its execution are separated by a gap that can be filled in only by an interpretation, and conversely, that a wish seems already to know what will or would satisfy it, just as a proposition or thought seems to know what would make it true -even when there is nothing there. It is striking that Wittgenstein must have decided that an investigation into the pictoriality or intentionality of the proposition, which the Tractatus presented as the mystery of negation (How can a proposition be false but meaningful?) can be more readily presented by considering the pictoriality or intentionality of orders and commands, wishes and intentions. But, again, too few clues are given to the reader.
The rest of the remarks of this "chapter" dwell on other aspects of pictoriality, e.g. that a wish, expectation, belief seem of their nature unsatisfied, that an expectation contains a picture of its own fulfilment, that satisfaction is a logical relation, not a psychological one. A conclusion is arrived at in §445: "it is in language that an expectation and its fulfilment make contact", i.e. not in a word/world relation, but in a grammatical nexus, a rule for the use of words: "the expectation that p" = "the expectation that is satisfied by its being the case that p".
It is, I trust, clear that this series of remarks constitute a deep criticism of the picture theory of the proposition. The Tractatus misconstrued an array of grammatical relations with a battery of metaphysical connections between language and reality. But the remarks are too compressed to be readily intelligible, and they are too well disguised for a reader, even a very intelligent reader, to realise that it is the Tractatus picture theory of meaning (advanced as a resolution to the problems of the pictoriality of the proposition) that is the true target. To be sure, its relocation would have helped the reader.

The ending
The final sequence of remarks in the book deals with the correct analysis of meaning something. The final two remarks, § §692-3, revert to the example of giving a pupil an order to expand an arithmetical series as it was left in § §186-90. For, as we have seen, one tempting dead-end is to hold (correctly) that the teacher meant the pupil to go "1000, 1002, 1004, …". But these two remarks add nothing to what was already clarified in § §186-90 and add nothing to the discussion of meaning something in § §661-91. The repetition seems redundant, and an unsatisfactory way to finish the book. To be sure, Wittgenstein may well have thought that he could just stop when he had said all he had to say. But that seems to me unpersuasive. A book needs an opening no less than a house needs an entrance. It needs a closure no less than a house needs a roof. It is a poor excuse for the absence of a roof to tell the client that one had run out of tiles. 8