 Okay good morning. So today I want to begin with compare and contrast between Putnam and Wittgenstein on meaning and then we'll go on to look at various aspects of how you might how to develop Wittgenstein's own picture. So here's a quotation from Putnam. Cut the pie anywhere you like, meanings just end in the head. Really that could have come straight out of Wittgenstein. Well I mean the style is quite different but he doesn't have Putnam's folksy charm but the message is very similar right? I mean when Wittgenstein is saying you're going to teach you're saying this is water that's water that's water that's water that's not water that's not water and the child has to get what the teacher is after, pull it together, grasp the meaning in a flash and the natural picture is that what goes on and that is that the child gets the right kind of image in their head and as a result of that can go on to say this is water that's water that's over there is water and so on. And Wittgenstein's point about this is the image does no work. All that happens is that the teacher gives all these examples and then after a bit the pupil works with examples in the same way as the teacher did. The idea of an image in the head it doesn't make any difference to what happens next. It can't, its meaning is only explained by what happens in concrete instances. And when you think about what Putnam is saying about earth and twin earth the message there is really the same. Water means something different on earth to what it means on twin earth. Though people are the Oscar on earth and twin Oscar on twin earth have exactly the same images in their minds. So what's giving meaning to the sign is not the image in the mind it's something about the context in which the images are occurring. That was a point about twin earth. It's also the point about brains and vats. If you think about what Putnam says in brains and vats he says let me see if I can remember something about an ant with a mental image of Winston Churchill wouldn't be representing trees. Let me try that again. Yeah if you've got a mental image of a tree that's not you representing trees just because you've got a mental image with the right pictorial properties. What makes it a representation of trees is the context that the image is figuring in or if the ant draws something in the sand that looks just like Winston Churchill. That doesn't make it a picture of Winston Churchill because what matters is the context in which the image was generated and there isn't the right kind of causal context for that to be representing Churchill. So Putnam's point there is really is there a difference between Putnam's point and Wittgenstein's. The negative point about the role of mental imagery in meaning seems to be exactly the same. Is that right? I mean the the style is so different but the message seems to be just the same. So then you might ask well what is it that Putnam thinks fixes reference and there seem to be two lines of thought in Putnam about what's fixing reference. One is the causal connections between you and the substances in your environment. What substance it is that you're encountering and responding to when you use the word water or whatever it is. And the other thing is for many terms like molybdenum or elm or beach you might not be able to differentiate these things very well yourself but you rely on the existence of experts in your community and they sort out for you what the word means. So with natural kinds the picture is that there's an underlying structure producing various observable characteristics of the thing. You pick up in those observable characteristics it's wet it's transparent it participates in the evaporation precipitation condensation precipitation cycle. It has only observable characteristics of water and you therefore use the sign and what you're doing there is responding to that underlying structure and in the case of a more arcane natural kind like a molybdenum or aluminum then it may be that what's going on is that experts pick up on those observable characteristics of the stuff. Experts give a name you just get the word from the expert you rely in the expert to give meaning to the sign. It happened to me a little while ago you ring someone up and a childish voice at the other end says daddies at a conference on low temperature physics and presumably you're not dealing there with a ten year old who knows what exactly what low temperature physics is but how should I say they are they are operating as a mouthpiece they pick up the word from other people and it would be perfectly possible to say no that's not right you know it's really very low temperature physics or high temperature, whatever. You can get that. Okay so platinum on the other hand doesn't have anything like Wittgenstein's notion of a custom. I can't see that there's anything there that maps on to Wittgenstein's notion of a custom so that raises the question whether you really need the notion of a custom. I can't these appeals to causation and the division of linguistic labor do the work that Wittgenstein is trying to make the notion of a custom do. I don't myself see that they can but the situation is quite complicated there might be something I'm missing. In division of linguistic labor first I really glossed over division of linguistic labor in the class because I can't see myself that it's so fundamental to our use of language and it's certainly true and important that when we use what swing you use terms like low temperature physics then there is a generally recognized body of experts on these things and you give away to them you defer to them as to what the right use of the term is but that just presupposes that for the experts whoever they are whatever the subject is for the experts there's going to be such a thing as going right or wrong you haven't explained what going right or wrong consistent for them all you've said is that you accept them as authoritative on whether you are going right or wrong in your use of the sign but there's still a more basic question whether it is for them to be going right or wrong in our use of the sign. So that leaves us with causal encounters and the basic problem with these causal encounters with a substance or a person or whatever it is that you're talking about is that it's very hard to see how causal encounters of themselves can be what generate the existence of standards of rightness and wrongness and just things colliding with one another things impacting on one another doesn't of itself mean that there's any right or wrong here so how can it be that substances or people impacting on my use of language can mean that there's such a thing as going right or wrong here. Gretzky's idea was we can see how causation becomes something normative something for which there are standards if you talk about the biological function of something being to causally indicate the presence of something else so if you've got a structure in your brain whose biological function is to indicate the presence of water then you can make sense of that biological structure getting it right or wrong when it causally responds to water but the basic problem there was that point that biological optimality is not the same thing as the optimality of truth that's to say what's do I need to go for that yes okay what's helpful for an organism what's adaptive for an organism might not be the same thing as getting it right if you've got an organism in a water poor environment let's suppose you got an organism in a water poor environment it might actually be quite adaptive for that organism to get false positives about the presence of water right it will it will keep it spirits up it will keep it trying even when there's nothing particular to try for it may help it in navigating around when these examples of thinking in more popular than you are thinking you have more control over what you're doing than you really do thinking the future is going to work out better than it does was biologically optimal what's adaptive is not necessarily getting it right or getting it wrong suppose you're a presidential candidate and you're wondering who's going to win the next election I mean presumably there's only one rational answer to have at the moment right one one of these guys should be thinking yes I'm going to win another guy should be thinking well looking at it looking at the facts the probability is I'm not going to win right but nobody for a moment expects that the candidate should really think like that what you really expect is that they're both thinking yes we're going to do it that's adaptive right that's all that's what's optimal that's what's normatively optimal so it's normatively optimal for an organism need not be truth was adaptively optimal for an organism need not be truth okay I mean presumably a similar point applies even to humble bacteria well that's what I mean about it yeah that some kind of optimism might be adapted for many many animals in many different situations so biological optimality look it looks like there's a difference between biological optimality and the optimality of truth and once you take that away and you try and do something more austere like for those causal asymmetry theory then it's very you're just left with a pattern of causal relations and it's very difficult to see how a mere pattern of causal relations you remember cause it cause a symmetry if the if the one causal connection hadn't existed the other wouldn't have sorry the high pressure low pressure exactly yeah did you do that again horses and cows and all that okay yeah you got enough of that okay so there you've got counterfactual relations between causal connections but that's all and how can a mere set of counterfactual connections among causal relations constitute the existence of standards of right and wrong so I think Putnam's I mean Putnam and Wittgenstein share the negative point about images but I think so far as I can see Putnam's appeals to causation and the vision of linguistic labor gave out on you so I think we have to look at Wittgenstein's notion of custom or practice and it looks like the idea of the notion of customer practice is going to explain how there can be right or wrong in your use of a word but then the next question is how does that work what is a customer practice how does that one of the mechanics how does that explain the existence of standards of rightness and wrongness in the use of language and also looking at Wittgenstein he never actually you have to acknowledge that he never actually frames it like that in terms of explanation so I just want to raise this there's a background issue here is that even the right question to be asking that may be the wrong question to be asking you the text okay okay so that's how I think Wittgenstein and Putnam connect strong agreement and a negative point and Wittgenstein's promising something more than Putnam can deliver on the positive picture are we all comfortable I don't mean physically are you mentally comfortable are you intellectually comfortable is it okay okay there's I just want to go over something I touched on fairly briefly last time this distinction between semantic thinking of meaning in terms of semantic value and thinking of meaning in terms of use because in the face of it that's a big contrast between Putnam and indeed everyone else we've looked at on one hand and Wittgenstein on the other so there's a general notion of semantic value the semantic value of a term is how that term contributes to determining the truth or falsity of any sentence containing it so for a name the semantic value of the name is what is a semantic value of a name yeah the object for object right the object it refers to yeah for he said yes what is a semantic value of a predicate it is a mapping from objects on to truth or falsity so if you've got a name if you got a sentence Raleigh smokes and you know what Raleigh refers to you know which object and you know what function from objects on to truth values smokes is a semantic value of smokes then you then you have determined the truth or falsity of Raleigh smokes yeah it will depend on whether that function maps this object on to truth yes okay this is really this is really so important I think this is the center of most many philosophical questions so I just want to take this quite slowly we had this last time suppose you say good or refers to good all and is human is true of an object just when that thing is human then the semantic value of not is given by the truth table okay this is your opportunity to pause if you think this is not quite clear that the reference the mapping from objects on to truth values and the truth table they're all parallel to one another and a little bit different because when you have a sentence like good or isn't human then the reference of good or in some sense is what you're talking about the semantic value of is human isn't usually what you're talking about though it can be the topic of the sentence the truth table of not is never what you're talking about you see what I mean so I mean if you wanted to talk about the truth table you'd have to get a name for the truth table and refer to it yeah just the mere fact that you're using the word not in a sentence does not mean that you're talking about a truth table so there are notion of what a term is about there is different to the notion of semantic value and these three things are all very different when it comes to when you ask questions like what is the sentence about but if you just get this abstract conception of semantic values how the term contributes to deterring the truth or falsity of a sentence containing it then that is then when you've got that very abstract conception that you see the parallel here this is really important is the parallel clear what could be clearer than that okay suppose you want to explain the meaning of and how would you explain the meaning of and a truth table yes very good okay so this is a little bit fancy right for not it's just two lines here is four lines I hope I've done this right I always make mistakes with the truth tables yeah that's right yes yes I say that's okay okay so if they're both true then A and B is true and if only one of them is true or they're both false then A and B is false yeah so that's a map from truth values to truth values and that gives you the semantic value of and now on the face of it that talk about semantic value is exactly what Wittgenstein is throwing out it's very natural to think that's how you know the meaning of and you can know the truth table yeah that's how you know the meaning of girdle or you know the meaning of his human you know the semantic value that would be the thing that tells you how to use the term in particular cases so that if you know what that map is from objects onto truth values for his human then when you're confronted with something when you confronted with a chair and asked is that human then what you do is you consult your knowledge of what the map is and then you see if that thing gets mapped onto truth or onto falsity and then you give your answer so the map here will be like the thing in the middle the thing that like the mental image it will be the thing that you measure the object against to see whether it satisfies the predicate yeah it will be the thing in your head the rule that you use to determine whether or not the word applies to the thing similarly your knowledge of good or the semantic value of good will good will refers to good will that will be the thing in the middle that when you're encounter something and you ask is that good will then you'll consult your knowledge of the reference of the sign to see if the thing in front of you is that reference and biggest nice point is that isn't that that notion of semantic value is the thing that does no work there's only your applications of the word in particular places so if you're explaining the meaning of a and b it would be better to forget about the truth table and just look at the rules of inference for a and b so what is the introduction rule for a and b what is the introduction rule for and as you get a and you could be you can draw the conclusion a and b I don't say it I don't say it's that deep an insight but um but that's what it is right and what is the elimination rule for a for for and if you get a and b then you get a and if you get a and b then you could be so that's a pattern of use for a and b and that's a different thing to the truth table and in Victor science picture it would be better to say you learn the meaning of and by learning the pattern of use you just do operate in this way in particular cases this just this in fact is how you do reason that's all a hard fact about the use of language if you look at what someone is doing in their use of and that describes what they do that describes their custom of using and and this thing is just idle if you forgot about this thing and said well let's just stick with the pattern of use you'd have all the facts so all that all that work you did in truth tables was really beside the point right because all that really uses the pattern of use so if you could so that clear that that will be what replaces the talk about truth tables in Victor science account yeah there's a chance to protest if this is not completely some of you look extremely puzzled yeah yeah that's right it's not it's not to say the truth table is wrong it's just to say it doesn't do any work in our understanding of the sign once you've got an understanding of and you might draw up a truth table but the great mistake is to think that the truth table is fundamental and you can derive your use of and from it yeah I mean actually in this case when we talk about adding one or water or these other examples we had or having an image of blue or having an image of a cube or whatever then in a way is relatively informal that you can't derive the use from that knowledge of semantic value but in this case there's really something like a proof of this because suppose you didn't have any patterns of use for logical signs okay suppose you you've got and or if then the basic logical signs and you don't know any of this pattern of use all you know is the truth table so suppose you've got truth tables for all your basic logical signs are you going to be able to derive the pattern of use from that it should be able to do it yeah how would that go right isn't that intuitive I agree that's really intuitive you should be able to think well if you get a true and be true then you get a and b so that will give you the introduction rule yeah and if you've got a and b true then you must have a true and you must have b true so since when you get a and b true you must have a true you can get the first elimination rule and since when you get a and b true you've got b true then you get the second elimination rule yeah so that I just derived the rules of inference from the semantic from the semantic value that's the way the classical picture says it should go yeah you know how to apply the term because you know the semantic value yeah you know how to apply the word girdle because you know what it stands for you know how to apply the word water because you know what it stands for but listen to what I did there I just said I said if you've got a and b true then you've got a true and you could be true so I just used some rules of inference I used the rules of inference for if then yeah and when I said and I said if you get a and b true you get a true and you could be true so you get a true that's using the elimination rule for and yeah and when you think about it there isn't any way around that because deriving is using inference rules yeah so that's using rules like this so in order to derive anything from this I've already going to be using the inference rules that's just a brick wall that is no way around that yeah so the most you could hope would be to fiddle about with which particular infant's rules you're using but in general if you take all the logical constants simultaneously and say I got all their truth tables if you took a child that was incapable of logical reasoning didn't have any of the principles of use for the logical of logical particles and you said there you go there's all the truth tables now get on with it you couldn't do it right yeah if that was your position you'd be absolutely helpless so in this case the biggest time general point can really just be proved you can't in general derive the pattern of use from the semantic value because the pattern of use is required to do the derivation yeah okay you can pause me at any point okay so when we're if we're going to replace the talk about semantic value here for goodl and for his human by something like this yeah then what we would need is something like instead of goodl refers to goodl you'd have something like in the circumstances say goodl is if when you're presented with a bespectacled sinister mathematician who proved the incompleteness of arithmetic say that's goodl right goodl is smoking a pipe but yeah that's just the basic rule and if you accept goodl is smoking a pipe you can conclude from that something like there must be a mathematician who proved the incompleteness of arithmetic smoking a pipe something like that so the inputs and outputs for your use of the sign will replace that talk about semantic value yeah so we really throw out everything we did in the first part of the course and I'm sorry I'm sorry about that there you go you course will throw all this out too no actually okay and similarly for is human you'd say is human instead of saying is human is true of an object just when that thing is human instead of specifying the semantic value for the sign you just say when you'll use it in these circumstances when confronted by this kind of thing say X is human and on the other hand the elimination rule if you accept X is human then conclude the circumstances must have been H however exactly you specify what should trigger your use of the term human I mean there is a question here about how you formulate these patterns of use you see it's very easy here to give a concise statement of what the pattern of use is it's not easy here to give a concise statement of what the pattern of use is but presumably I'm not sure about this but presumably in both cases something like Wittgenstein's general point will apply that your understanding of what this pattern of use is has to be explained by reference to what you do in particular cases that's to say if you were had to explain to someone what these general schematic things mean you'd have to say well for example if I say Raleigh smokes and Isaac fishes then I'll conclude Isaac fishes or if I know Isaac fishes and Raleigh smokes I'll conclude that Isaac fishes and Raleigh smokes so you just have to give examples and that will be true presumably for these cases too you just give examples to explain what the general schematism means so presumably these formulations of patterns of use will be subject to Wittgenstein's point that they can only be explained and understood by giving examples. Here ends the second lesson. That was a yes that's right. You're not meant to be just writing down my answers though. My task is to formulate the problems. Your task is to sort it all out. Did I get it wrong after all? No, no, no. I'm just wondering about someone who says that all we actually need is understanding. According to... That's right, yeah. Well I don't really see how you're going to do the derivation. I mean there's got to be something important and right about the idea that the truth table shows how these rules of inference are proof-preserving. The truth table just spells that out for you line by line how this trunc and I mean this is the simplest example you could possibly have but for any even much more complex examples the same general point is if you get truth at the inputs you get truth at the outputs. But Wittgenstein really throws out that insight it seems to me he doesn't leave you any way of acknowledging that that's an insight and the argument is if you just say schematically I want my rules of inference to preserve truth yeah in a particular case you've got to do the explanation for why these rules of inference preserve truth and in giving the derivation in giving the explanation you will actually have to use lots of principles of inference. I mean the simplest thing to begin with is you're going to have to say if you have A and if you have B then you can conclude A and B. So going back to the truth table you'll have to say if you have A and if you have B well let's look here I've got these both true then from the truth table see all that if then reasoning so you're just using the rules of inference for if then yeah. I mean another way to put it is you know the story of Lewis Carroll's the what is it the kill using the tortoise yeah so Achilles in the tortoise run a race and Achilles crosses the line first and Achilles turns to the tortoise and says look I won and the tortoise says no I don't see that you didn't win and Achilles says look I crossed the line first and the tortoise says yes that's right and Achilles says well if I crossed the line first then I won and the tortoise says yes that's right and Achilles says so I won and the tortoise says no I don't really see that. And so Achilles says, look, here's the truth table. I'm paraphrasing a little bit, but Achilles says, look, here's the truth table. Look, so you accept the truth table for ever then. Picture, if you will, the truth table for ever then. And Achilles says, look, I crossed the line first, right? And is true also true? And the trotter says yes. And Achilles says, so it's also true that if I crossed the line first, then I won. And the trotter says yes. And Achilles says, and look, here's the truth table. And the trotter says yes. And Achilles works through all the lines. And the trotter says yes to all of them. And then Achilles says, so I won. And the trotter says, no, I don't see that yet. And the point is that if you don't get the rule of inference, then no amount of assenting to the truth table, Achilles keeps saying things like, well, look, logic will take you by the throat and force you to accept the conclusion. And the trouble is it really doesn't. If you think of logic as a truth table, unless you already get the rules of inference, the truth table can make you do nothing. The truth table is silent, really. Yeah? Yeah, I mean, related question, do you think the pattern of use is basic, or not taking the path of use as basic? Well, I'll take you to be prior to explaining the truth conditions. That's what I mean by taking the pattern of use as basic. Let's talk about truth in Wittgenstein's picture. As I read him anyhow, I mean, all this is up for negotiation. But as I read Wittgenstein, his picture of truth is the notion of truth is that it's completely peripheral to any discussion of understanding and meaning. If you want to say what truth means, it means something like this. You get a sentence P, then P is true. It's the same thing as saying P. And that's it. It's not an explanatory notion. It's a best notion you might use when you can't be bothered repeating the whole of P. Yeah? As someone makes a comment and you say, yeah, what he said, you can't be bothered going through the whole comment yourself. Yeah? Or if someone makes a comment and you say that's true, it's just you can't be bothered repeating the whole thing. It's a useful shorthand term, but that's all truth is, yeah, there's nothing fundamental about it from the point of view of understanding and meaning in this picture. So that's the same thing as throwing out the truth tables as basic to an understanding of the signs. Saying that A is true is just another way of saying A. Yeah? Okay. So what's a custom then? If we've got the patterns of use and we want to say the patterns of use, you can go right or wrong in your use of a sign, then what do we get from making shine on what a custom is? I'm just going to, I think all I can do really is give you a sense of what he means by custom. I think it's very tantalizing. It is not a properly spelled out notion. If language is to be a means of communication, there must be agreement not only in definitions, but also queer as this may sound in judgments. And you'd naturally think with language, like all that matters is that we all define our terms in the same way. We don't have to agree about any particular case, so long as our definitions of the terms are all the same. And his point is, if you think about something like the example add one, we've really all got to agree in the particular examples before we can be talking about taking the successor of a number at all. You couldn't make sense of everybody agreeing on the definition of add one, but disagreeing about whether three comes after four or four comes after five. Your understanding of add one only comes because we have all our agreement on the particular cases. You naturally think of semantic value, as someone just said, giving a foundation for your use of the term. And here's the general negative point here. How am I able to obey a rule? If this is not a question about causes, then it's about the justification for my following the rule in the way I do. And if I've exhausted the justification, then I've reached bedrock and my spade is turned. When you ask why do you derive a from a and b, you can't appeal to a truth table to explain why you do that. You just do it. The best you can say is this is simply what I do or you might say this is the way of my people. And that's all you can say is to justification. Justification stops at a certain point. It stops sooner than you might have expected. I have, well, yeah. It is when I obey a rule, I don't choose, I obey the rule blindly. And the idea of, how should I say cited following of the rule is that you see the thing, you see what kind of thing it is, you match that up to your knowledge of what the term stands for. And consequently, you apply the term. That's a natural way to think of working with a word like red. You know what red is? You got the image in your head. You see the object you ask, you see it and you say is that red? And then you match it up to the image in your head. And if it matches, then you say, yes, that's red. And the point is to throw out that kind of image of what's going on when you say that something is red. All that happens is you see the thing and you say that's red. There was no vision, there's a sense to which vision didn't really come into it. We talked about these passages a little bit last time is what we call obeying a rule, something that would be possible for only one man to do and only once in his life. Well, no, because you don't have a background of definition or semantic value. You will only get a pattern of use. And it doesn't make sense to talk about a pattern of use being exhibited just once. As your time is getting short, so I'm going to, when you think about how he's using the notion of custom, it connects up to something like our form of life to ways of using terms that we could pick up from each other part of what Kavel called our will of organism. So it's natural to think, well, what is that? How do we make more explicit what he means by the world of organism within which we have customs or practices to obey a rule to make a report to give an order to play a game of chess or customs uses institutions to understand a language understand a sentence means to understand the language to understand a language means to be master of a technique. So the kind of analogy you might have here would be something like a carpenter using a chisel is not that you need to have in your head a good carpenter needs to have in their head a whole bunch of rules for using a chisel is that using a chisel is part of mastery of a whole bunch of tools. And the ability to use those tools isn't to be explained in terms of you being able to say or having some images as to what you do. It just is being able to use them well in context. There are two really tantalizing examples here. One is think about this case. It is of course imaginable that two people belonging to a tribe unacquainted with games should sit at a chess board and go through the moves of a game of chess and even with all the appropriate mental accompaniments. So you come across these two guys in a clearing in the jungle and they're at a chess board and they're making or they come from a people that just don't have games and they're going through all the motions of a game of chess. And they even has he says of all the mental accompaniments. One of them is saying yes yes he's on the run. The other one is saying oh no boy am I in trouble. They have all the mental accompaniments. If you were to see it if you just wandered in upon this scene you would say they're playing chess. But I take it the point here is the lack of a context means that they couldn't be correctly described as playing chess. This is not part of a general custom what they're doing here. And a way to see what is wrong with taking it to be part of a general custom is to say and imagine a game of chess translated according to certain rules into a series of actions which we do not ordinarily associate with a game. Say into yells and stamping of feet. You could do that. Right. You could just translate the game each move in the game into a particular kind of yell a particular kind of stamping of your foot. And I suppose that what happens is you come upon two people in the clearing yelling and stamping instead of playing chess. And you say look I've got a video I can do an analysis of what is happening here. And I can translate this all into the moves of a game of chess. It's true these these guys don't have games in general. They are much more serious than that. But I can show you I can translate this into a game of chess. Nobody would take this seriously for a second. Yeah. The mere fact you could do that that you've come upon this shouting and stamping. And you said I can translate that into chess. That would make you a tall inclined to say these guys are actually playing chess. So if you come upon people making the moves in the board the translation into saying this is chess is simpler but it's no more justified to say this is chess. And one reason for saying it's not chess is you couldn't pick up from these chat from these stamps and yells how other people other humans could not pick up from these guys how to go on with the stamping and yelling how to have further games of chess in terms of stamping and yelling. You've got something that with chess we can pick it up from each other. We know how to go on after a bit just as with learning add one when you get a certain number of examples you know how to go on. You can do the thing yourself. These guys if they're stamping and shouting you couldn't pick up how to go on from them. If humans actually did did not try if you had someone who said let's not play regular chess. Let's just do it in terms of stamping and shouting. That would not be something that you could teach other humans in a way that we teach ordinary chess. People couldn't pick it up and get the hang of it in the same way. Imagine someone using a line as a rule in the following way. I think what he's got in mind here is something like this. You will line here and you're a pair of compasses by which it's something like this. Is this right. You have to tell me I think this is what a compass is. It's going to pair of pencil here and a pencil and a spike here. Is that right. That's a compass. Yes. So you put the spike along here and you've got the pencil along here. And someone says look this is what I'm doing. This is the way of my people. Here is my way of going on with a particular example. So he holds a pair of compasses. Carries one of its point along the line while the other one is drawing the line that follows the rule. So this is like the rule governing your use of the language that line along there. And the compass is like what line is going is like your use of the word in particular cases. But let's suppose while he moves along the ruling line he alters the opening of the compasses apparently with great precision. Looking at the rule the whole time as if it determined what he did. So you've got someone who's very carefully going along going like this. Yeah. And you and you say yes look at that. Look at that. He's muttering to himself as he does it. And you say I can't see what you're doing there. Watching him we see no kind of regularity in the opening and shutting of compasses. Or we cannot learn his way of following the line from it. Here perhaps one really would say he thinks he's got a rule here. The original seems to him to intimate which way he's to go. But it's not a rule. So a custom is not something where you just have a lot of uses of words in particular cases. And we've given up the idea that you know the semantic value of the term. So you just have uses of the word in particular cases. This kind of thing where someone uses the word in particular cases but you can't pick up from them how to go on. That doesn't count as a custom. If instead of saying if I say zero one two three four five six after a bit you can pick up from me how to go on. If I shout ninety seven three four hundred twenty one eighty two. I can do this for quite a while and at no point are you going to be able to say I know I get it. Now I can go or rather you might but what you've got was you could do anything you liked next. As opposed to zero one two three or this is red that's blue and so on. When you can pick it up. So something about a custom is not just that you've got uses of the term in particular cases and no semantic foundation. It's uses of the word in particular cases and there is such a thing as getting the hang of what someone else is doing getting the gist of it so that the thing can be communicated so that other people can learn from you what you're doing. You need something like that to have a custom. On Monday we'll look at Kripke's interpretation of Wittgenstein on all this.