On the Benefits of Introducing Undergraduates to the History of Mathematics—A French Perspective Anne Michel-Pajus IREM PARIS VII" Some aspects of the reform of the preparatory levels France has a centralized education system that officially prescribes the various courses of instruction that students follow. In the system I teach in. some $0,000 students un- {ergo their frst two years of tertiary education. At the end of this time, the great majority of them take competitive ‘examinations t0 gain entry to engineering schools: some others enter the Ecoles Normales Supérieures in order to become researchers and teachers at either tertiary or sec ondary level. In mathematics, the same teacher teaches one class 16-20 hours a week. The basic class of about 45 students cither works as a whole group or is divided into subgroups according to the activity. For two (0 five hours a week. the students work on exercises in groups of fen fo twenty. with or without @ computer, One hour every 0wo weeks they work in groups of three for oral questions. “This system has just undergone an important reform, affecting both its structure and its programs of instruction, ‘This reform aims to reduce the importance of mathematics iaself by relating it more closely to physies and engineering science, It also aims (0 develop a spirit of initiative in the students ‘The mathematics syllabus is in two parts. with accom: panying comments, The first sets out general educational ‘ims while the other deals with the topics of linear algebra. calculus, and geometry. The students must know how 10 use both calculators and programs that perform symbolic ma- nipulations (Maple, in my institution), There are differences in content according to the major followed (mathematics and physics. physics and engineering sciences. physies and chemistry. physies and technology) but the educational ob- lectives are the same. In sum, mathematical education must multaneously develop intuition. imagination, reasoning. and rigor.” A\ short addition concems the history of mathematics: It is important that the cultural content of mathematics should not be simply sacrificed to its technical aspects. In ‘particular, historical texts and references allow the analysis cof the interaction between mathematical problems and the construction of concepts. and bring (0 the fore the central role played by scientific questioning in the theoretical de- velopment of mathematics. Moreover. they show that the evolution and that dogmatism is not advisable, ‘Another innovation isthe introduction of project work. ‘The history of mathematics is also mentioned in relation to this; “The study of a subject brings an increasing depth ‘of theoretical understanding together with experimental as- ”18 Using History o Teach Mathematics: An Imernational Perspective pects and applications as well as the application of com- pputing methods. It may include an historical dimension In the first year, students choose their project freely. In their second year. they must fix their area and subject in a very wide framework (for example. dynamical systems in mathematics, measurement in physics). At their final ss- sessment, each student has to present a page long summary and to speak for twenty minutes about their work before two examiners, ‘The evaluation deals mainly with the communica tion abilities of the students, but for the teachers who are constrained by a heavy, prescriptive program, this project. Which takes up two hours a week, allows the freedom for them to introduce the history of mathematics. These reforms began in 1995, and so Ihave only a lim- ited experience to draw upon in writing this article. How- ‘ever, [have found working in the area of the history of ‘mathematics most enriching, and if my readers can glean some ideas from it, its purpose will have been achieved, ‘Texts available in France The first difficulty isto find documents students can work fon, One of our basic principles for students undertaking such a project is that any historical study must involve ‘at least one extract from an original text, But since it is usually difficult wo read them, secondary studies need also to be made available. The majority of useful documents in France have been published by the Commission Inter IREM d'Histoire et Epistémologie. IREM stands forthe In- stitut de Recherche sur Enseignement des Mathematiques nstiture for Research in the Teaching of Mathematics In France these institutes bring both secondary and tertiary teachers together. Those teachers working on the applica- tion of the history of mathematics in the regional IREMs ‘come together nationally three times a year. These meetings comprise lectures and workshops and allow experiences to be exchanged. Proceedings and many other documents sre published. Among these are the “problem-documents” in Which explanations and questions are linked to selected passages of an historical text.? The collection provides homework for students in various classes. An example of such a document is attached (Appendix 1), A selection has been similarly translated and published by The Britis “Mathematical Association.® This network is also commit- {ed to publishing texts—sources which are difficult to find in French in any other way. As a result I have participated in the edition of 4 History of Algorithms which systemat- ically presents the historical texts translated if necessary situated in their cultural context and accompanied by math ‘ematical explanations. The majority of the texts to which I refer here are drawn from this work.? ‘Why choose approximating algorithms? Apart from the fact that my work in this area has allowed re 10 collect large number of documents. this choice presents many advantages, © the same topic allows one to cross centuries and civiliza tions, and to meet many problems which are still open For example, students are always fascinated by research into the decimal expansion of = + The question under consideration is clear: itis easy for students to compare the efficiency of procedures. which they can even est using a computer. and thus see the use- fulness of theoretical concepts whose depth and general- ity they otherwise find difficult to evaluate for the lack fof 4 theoretical overview. For example. iterative meth~ cds of solution of linear systems involve the topology of spaces of matrces:* The interdisciplinary aspect ofthis theme is wort noting, as the fist studies arose from astronomy and surveying. + Students can see how technical necessities drive the con- struction of new concepts and sometimes inspire them. Consequently. there is no boundary between pure and applied matheratics—the noble mathematies of mathe- ‘aticians and that of enginces. + Being abe to write an algorithm ina computer language, such as Maple or Mathematica, involves the effort of freeing oneself from the restrictions of one specific lan- guage and from a particular set of ideas. This is similar to the effort needed to understand ahistorical tex. There are. of course. other advantages tied up with course work, such as the connection with the theme of dynamical systems. Some examples 1. A Project: The approximation of =. Six students chose to divide this topic up among themselves for their project. To assist them in understanding the sources 1 fur- nished them with problem-documents put together by sec ondary teachers for their students. They obviously obtained other general references concerning the history of mathe- ‘matics. An example is piven as Appendix 1: it isa problem- document relating to a text of Euler, which was originally intended for students in their final year of secondary school. [As part of the task of carrying out this project. I in- sisted that they state precisely the problems under consid geration. the concepts, and the tools in use atthe period of| the text they were studying by asking them to reflect on the following questions What isthe author seeking to determine—a number, an anes, «surface? What is he seeking to prove? + Why do it? ‘+ What fools—notation, theories, theorems—are available to him? What is implicit? + Dees it seem rigorous to you? “The texts which the students chose to present 10 their class were: ‘Archimedes®: Measurement ofa circle (Ist century BCE) Descartes: De la quadrature du cercle (1701) [Concerning the quadrature of the circle] Leibniz!: Lee a La Roque (circa 1780) [Lewer 10 la Rogue! f Euler: Des quantites transcendamtes gui naissent du eer cle (1748) [Transeendant Quantities arising ftom the Circle) Richardson: The deferred approach ofthe limit (1927) With reference to the historical aspects: + students are astonished to see how slowly the idea that 1 was actualy @ number arose. Until the seventeenth century 7 was never writen, or even thought of. as a sumber in formula. Inscad we find a method of cal culation, writen in the language ofthe time, which can be used to caleulate the area or perimeter of a circle (This can be seen, not ony in the works of Archimedes. ‘but also in the work of the Chinese). Unhappily the text which is available to us is not very enticing. while st- lens ate regaled with the geometry of the triangle used by Archimedes, going so far as 10 find four different proofs of one theorem. + they are surprised t0 see in 1927. an almost contempo- rary mathematician, Richardson, picking up the example of Archimedes in order to ilustrate his method of ex- trapolation. Since this isa little known work, an extract is attached (Appendix 2) « they have determined that there are several stages inthe development of approximation. «frst stage: the basi idea of en approximate value (in- teresting for its technical uses) «second stage: the determination of an interval of va- ‘grasp. Generally, even in their exercises they give nu merical results which show all the numerals which appear on their calculator display. without troubling themselves with the precision of the procedure they are using, even when the objective of the problem is Introducing Undergraduates tothe History of Mathematics 19 to evaluate this procedure!) The text of Archimedes illustrates this step in a remarkable manner. ‘¢ third stage: practical methods which provide more and ‘more rapid approximations such as the use of series continued fractions, and so on. The region of validity is sometimes made explicit by giving an inequality as in the works of Wallis or Huygens, for example. and is sometimes implied, as in the work of Leibniz and Euler who give an alternating series, (Leibniz first es- tablished that the partial sums are alternatively above and below the sum to infinity.) # concems as to how to speed up the rate convergence lead to theoretical methods of the acceleration of con- vergence which are useful in other contexts. One ©x- ample, which has the advantage of being recent. is found in the article by Richardson «The concepts of infinite series. convergence. and rate of convergence are being built up concurrently with mathematicians carrying out theoretical studies as 10 the nature of = and its relation to the quadrature of the circle 2. A computer-related study: Newton's method. This is equally well known under the names of “the tangent ‘method or “Newton-Raphson” even though neither New~ ton nor Raphson spoke of tangents or even of geometry in this context. Moreover. a century passed before Lagrange saw that they all came down to the same method. The algo- rithm put forward by Newton is very clear and easy to put into effect with symbolic manipulation software. (A Maple program is appended.). Proposed before the invention of Aluxions. it does not use derivatives. but only the idea that some quantities are negligible with respect to others. The principle of a calculus of approximations is not made ex- plicit in the form of a recurrence relation, unlike chat which Appears in Raphson. However, the writings of Raplison are more difficult to read and a much clearer explanation ean bbe found in Euler. These authors treat particular cases in which a recurrence relation appears. although they do not use the derivative explicitly in their writing. ‘The task is for students to enter into the logical pro~ ‘cesses of a seventeenth or eighteenth century mathemati- cian and to translate their methods into their own program ‘ming language, I make the further demand that they justify derivative. This historical aspect reinforces the idea thatthe derivative allows one to make a first order approximation, of that it allows one to replace a function by its fist order expansion—the tangent giving a geometrical illustrat his (cf. Appendix 3) of20 Using History to Teach Mathematics: A Inter 3. Commentaries associated with the course: The Cauchy-Lipschitz Theorem. I have searched for exam: ples in which techniques of approximating solutions have led to the theoretical proof of a theorem. The topic of dif ‘erential equations provides a remarkable example of this. ‘but one which lies outside the technical capabilites of my students. I can only refer to it in the course, The syllabus involves stating the Cauchy-Lipshitz ‘Theorem on the existence of a unique solution to the prob: lem of Cauchy. without proof. It also includes research on approximating solutions to differential equations by Euler's method. Now the first proof gi 1824. which remained unedited until its recent rediscov- ery. begins precisely with the approximation furnished by Euler's method applied to the interval (20. X| for a given partition. By considering upper bounds on the error. Cauchy showed that, under very general assumptions, the approx imate value tends toward a limit which only depends on X a8 the steps of the subdivision tend to zero. It then ‘only remained to prove thatthe function defined in making X vary across a neighborhood of 29 is. in fact, the solu- tion of the given differential equation, Picard’s work, which completes this theory, also refers back to Euler's meth although the object of the successive approximations is no longer the value of a function ata given point (as in Euler) bbut a function defined on an interval.?° Conclusion | would like to conclude by revisiting the educational as- pects of my syllabus. I have tied to “stimulate intuition ‘and imagination” by offering multiple viewpoints of the same mathematical object: “struggle against dogmatism’ by demonstrating that the process of mathematical creation {is cumulative but not linear and that each era constructs its ‘methods. concepts and proofs with a rigour appropriate to its intellectual framework: and finally show that there is no hierarchy between “pure” and “applied” mathematics, the= ‘retical concepts and their applications, and that even for future engineers mathematics is nox merely a “serviceable discipline” but also a marvelous success of human thought Tam not able 10 say 10 what degree. greater or lesser. have been able to attain these objectives. I do know that al Perspective this work has not only given me pleasure. but that it has also given some of my students pleasure as well. And that’ quite something. Endnotes * I would like to thank Stuart Laird (Rangitoto College. New Zealand) for his translation. sor his pertinent comments and for his careful rereading ofthis article and for his New Zealand warmth 50 welcome inthis winty season, * Problem documents for 15/16 years olin Brochure M:A.H. 2° 61. IREM PARIS VIL. la mesure du cele, pp. 8-13: in Brochure MATH, n° 79, IREM PARIS VIL. wos fiches sur le caleul de Pi. pp. 30-45; la mesure du ceele. p. 46: in History in the mathe: ‘matics classroom. The Mathematical Association, 1990. Reading Archimedes! messurement of a Citcle, Manne Buble, pp. 43-58 2 Histor in the Marhemanics Classroom, edited by John Fauvel The Mathematical Association, 1990, * Histoire a’Atgorithmes, J.L. Chabert a. Belin, Pais, 1994 History of Algorthms. .L, Chabert eta trans. by Chris Weeks Springer-Verlag. New York. 1999 2 The texts of Gauss. Jacobi. Seidel et Nekrasoy are to be found in History of Aleortims. chapter 9. ‘The English translation of T.L. Heath. in The Works of Archimedes. Cambridge. Cambridge University Press. 1997. x \ery far removed from the Greek text. much more so than the French translation of Paul Ver Eecke, Les oeunres completes \Archimede. 1921, ree. Blanchard. Paris, 1961. pp. L30-134. English translation and commentaries can be found i History of Aleortims. pp. 130-145, " Circuli_quadatio, excerta ex MS.R. Des Canes. Ed Ams- ‘erdam. 1701. in Les vewves de Descartes. published by CA P. Tannery. .X. 1908. pp. 305-305: re-edited Vrin, Paris, 1974 Translation and comments}. Chater in History af Algovithos pp. 153-156, * Leitmiz. Lene a La Rogue, Mathematische Schriften. pp. 88-92. Olms. Hildesheim. 1962: History of algorithms. pp. 138- \61. MF. Jozeau, M. Halle. M1 Buhler, Senes et quadeatures chez Leibniz. Histoire dinfini. IREM de Brest. 1994, pp. 273 mT. * Euler. Inroduction 4 Vanalse infintesimale, Livee 1. Ch VIL Des quanttés ranscendants gut naissent du cercle. English rans lation: J.D. Blanton. dneroduction tothe analssis of the Infinite NY. Springer. 1988, 1990. " Lewis Richandson, Philosophical Transactions ofthe Roval So ciety of London. Series 8, Nol 226. 1927, pp. 300-305, "© of. History of Aleoritims, pp. 378-381Iniroducing Undergraduate othe History of Mathematics 21 Appendix 1 Euler and the calculation of =* A. Preliminary Activities Deine Fon Ry £0) = fp tgée 1 2. a) Show that: Wt € R, za the existence and differentibility of F. Calculate F(e) and determine whether F is increasing or decreasing, a iaate ati ase ive b) Hence show that F(t) = P(t) + R(¢) where P is a polynomial of degree 7 and R(t) [tae ©) Show that vt € [01,0 < RQ) < 3, Define ig on =] 2) = Piven) 2) Justdy that Gis afferesable and calculate G'(2) for sin £ ») Catedate Gio). «) From the preceding deuce hat 4, a) Read lines | to 2 ofthe adjined text ©) What is the relationship between this text andthe preceding questions? 6) Explain the &e (= eteter) line 2, 5. a) Calculate F(1) and F(1/ V3). 6) Read lines 2 to 4 of the tox ) What is meant by the word series"? 1.G(2) B. Approximation of = |. a) Caleulate P(1) (see 2 above) to 2 decimal places. How close an approximation to 1 does this value give? Explain the &e of line 3. ) Read lines 5 to 15 of the text. Make explicit Euler’ computation, by calculating each square root to four decimal places. How close an approximation 10 7 does this value give? Improving the approximation a) Read lines 16 to 20, Verify Eulers calculations. In particular, prove the formula giving tan(a + 6). ») Suppose tana =f. Calculate tan ( where a+b a $), What ae the values of F() and F(})? Calculate P(3) and P(1) to 6 decimal places. Give upper bounds for (3) and (3). ©) Read the end of the text. Explain, using the notation of the problem, the equality of line 23. How close an approximation to = do you now have? Compare this with your earlier values. Euler, Introduction to Infinitesimal Analysis (1748) Thus set tang = t $0 that ¢ is the are whose tangent is t, and which we designate thus: Atangt, giving ¢ = A tang? Knowing the tangent, the corresponding arc will be ¢ = £-$+"—5+£—Wc. Then, assuming the tangent tis equal to the radius 1, the are ¢ becomes equal to the arc of 48° or 2, and we find 7 =1—!=+1—1-+ &c: the series which Leibniz frst gave to express the value of the citcumference of the circle. But in ofder to obtain the are length of a circle quickly, by means of such a series, it is clear that itis necessary to take a sufficiently small fraction forthe value of the tangent ¢. In this way, withthe help ofthis series, itis easy to find the arc length C, whose tangent t = as this arc will * M. Buhler Une approximation de pl Mnemosyne a? 40, IREM PARIS Vi. pp. 67-718 decimals. But the measure of such an th ofthe circumference as itis not possible to assign a relationship between ar nh angen = t and the entre circumference. This is why. inthis research, iis neceseay tothe me ae eich an aliquot part of the circumference and whose tangent is sufficiently small and can be conveniently expressed. To fulfil” this aim a 30° arc whose tangent = 1/3 is normally chosen, as ‘tangents of smaller arcs, which have a measurable relationship with the circumference, are to0 irrational. Thus, as the are of 30° / V3. we will have that 283 av3_ ava 1 33” 5.3? — 737 Cand fs And i by means of his Sess and an incredie amount of work, that te gol offing te vale of which wu ve above can be aint Tis alaation i all the more laborious a ll the tems ar iain and ech tm barely le than atid o is een ot: his ava an be eid fliwe sil chame steel Arar, ae he Sth ae represen ya bry conegen sees = 1h FE ve etn tego oe ce bea T-aagaiaa ivided into two ares a and b. such that a+b ‘Then since tang(a +) 5 wwe will have 1 tanga Ttangatangb=tanga~tangh and tang L= tanga Fangavang) = tanga ~ tang a) tad Non se lanka = 4: we vil find tang = J: then the vo arcs @ and & can be expesid by a rational series much moee onvergen than the preceding. and their sum will give the value ofthe are # The In tis way. by using the series we have previously given, the length ofthe semi-crcumferene can be found far ‘ore readily than it has been, Appendix 2 {wis Richardson. The defered approsch of the limit. Philosophical Transactions ofthe Roval Sociers of London, Series A, vol 226. 1927. pp, 300-305, {avious problems concerning infinitely many. infinitely small. pars had been solved before te infinitesimal calculus problen a nh the id things befor the calculus was invented. postponing the passage to the leit ener ake Problem had been solved fora moderate number of moderately smal ference. | se nPalton was sacovered by a hn fom thoy fllowed by antmeilexperinns, which ave pleasing results... ‘ Imagine that we are back in the time of ARCHIMEDES. aaa obvious very crude. approximation. take the perimeter of an inscribed square ~ 4V3 = 5.6568. As @ ‘evond approximation, ake the perimeter ofan inscribed hexagon = 6 exactly reo eco iste two estimates shoul be to one another as +: dy that is , fhe ero is proprtinal to the ‘avate of the coordinate difference. Thus the extrapolated value is 6 4(6 ~ 5.6568) = 6.7 a Othe extrapolate vale is ths ony 1/38 ofthe enor in the beter ofthe wo values fom which twos ours st tapeltion seems aust process, To gt as good a est fom a single nerd oean polygon it would need 10 have 35 sides. @Introducing Undergraduates tothe History of Mathemat Appendix 3: Newton , Euler and Maple 4a) extract of “The method of fluxions.” The mathematical papers of Isaac Newton, vol Ill, Whiteside ed. Cambridge University Press, 1969. pp. 43-47, [Written between 1664 and 1671, published in 1736, and explained by Wallis his Treatise of Algebra in 1685.) ‘When, however, affected equations are proposed. the manner in which their roots might be reduced to this sort of series should be more closely explained, the more so since their doctrine, as hitherto expounded by mathematicians in numerical cases. is delivered in a roundabout way (and indeed with the introduction of superfluous operations) and in consequence ought not to be brought in to illustrate the procedure in species. In the frst place. then. 1 will discuss the numerical resolution of affected equations briefly but comprehensively. and subsequently explain the algebraical equivalent in similar fashion Let the equation y? ~ 2y 5 = 0 be proposed for solution and let the number 2 be found. one way oF another, which differs from the required root by less than its tenth part. [ then set 2 +p = y and in place of y in the equation I substitute 2+p. From ths there arses the new equation p? Gp? + 10p—1 = 0. whose root p is to be sought for addition to the quotign. Specifically. (when p? + Gp? is neglected because of its smallness! we have 10p—1 = 0.or p= 0.1 narrowly approximates the truth, Accordingly. I write 0). 1 in the quotient and. supposing 0.1 — y = p. [ substitute this Fictitious value for it as before. There results Sy? ~ 11.289+0.061 = 1. And since 11.23 ~0.061 = 0 closely approaches ihe truth, in other words very nearly q = —0.0054 (by dividing 0.061 by 11.23, that is. until there are ‘obtained as many figures as places which. excluding the bounding ones. lie between the first figures ofthis quotient and of the principat one—here. for instance. there are two between 2 and 0.005), I write ~0..0054 in the lower part of the quotient seeing that itis negative and then, supposing ~0.0054- r equal to q. I substitute this value as previously. And in this way I extend the operation at pleasure after the manner of the diagram appended. {+2:10000000 (Zocoasi 0935548 [my] Tey. Dissen sp [soon soa F0aF ae +o [soos tie! 46 +10 +1 10 Teal | oon pinaay +e Tooceara ya | —oe0000NLsTEE +O O0OpRIERY —BarKIO +I Fooolwsie “ons TET Tags +1123 b) extract of Euler. Elements of Algebra, Chapter XVI. Of the Resolution of Equations by Approximations. §784-789 84, When the roots of an equation are not rational. and can only be expressed by radical quantities. or when we determining ther values by approximation: that isto say. by methods which are continually bringing us nearer tothe true value tila last the error being very small, it may be neglected, Different methods ofthis kind have been proposed, the chief of which we shall explain. 785. The fist method Which We shall mention supposes that We have already determined. with tolerable exactness, the value of onc rot: that we know, for example tht such vale exceeds 4. and tha it i less than S. In this case, if24 Using History to Teach Mathematics: An Internat Perspective wwe suppose this value = 4 + p. we are certain dhat p expresses a fraction, Now. as p is a fraction. and consequently less than unity, the square of p. its cube. and, in genera. all the higher powers of p, will be much less with respect 10 unity: and. for this reason. since we require only an approximation, they may be neglected in the calculation. When we have, therefore, nearly determined the fraction p. we shall know more exactly the root 4 +p: from that we proceed to determine ‘new value still more exact, and continue the same process till We come as near the truth as we desire." 786, We shall illustrate this method first by an easy example, requiring by approximation the root of the equation 0. Here we perceive, that x is greater than 4, and less than 5: making, therefore, 2 = 4 +p, we shall have 2? Must be very small, we shall neglect it, in order that we may have only the equation 16 + 8+ p* = 20; but as 16+ 8p = 20, or 8p = 4. This gives p = +. and x= 45. which already approaches nearer the true root. IF therefore, we now suppose 2 = 44+ pi: we are sure that p/ expresses a fraction much smaller than before, and that we may neglect with great propriety. We have. therefore, 2* = 204 + 9p’ = 20. or Up’ = —1: and consequently. 7’ = —: therefore raah-gaad. ‘And if we wished to approximate still nearer to the true value. we must make 2 = 442 + p!, and should thus have 8 = 20g + Bf” = 20: so that Shp” = — hy, of $22p" = — 3, = — 2. and therefore 2 = 4 = 39) value wii is so near the truth that we may consider the err as of no impomance ‘787. Now. in order to generalise what we have here laid down. let us suppose the given equation to be 2? = a, and that we previously know 2 to be greater than n, but less than n= 1. If we now make z = n+ p, p must be a fraction. and p* may ‘be neglected as @ very small quantity, so that we shall have i: or Inp = a ~ n?, and p= 2": consequently, 2 n? + 2np = oh hn Now. if n approximated towards the true value. this new value will approximate much nearer; and, by substituting it for n, we shall find the result much nearer the truth; that is, We shall obtain a new value. which may again bbe substituted, in order to approach still nearer: and the same operation may be continued as Iong as we please. For example, et 2* = 2: that is to say let the square root of 2 be required: and as we already know a value sufficiently In near. which is expressed by n. we shall have a still nearer value of the root expressed by Let, therefore, 1. n= A. and we shall have = and we shall have = = and we shall have 2 Bn This last value approaches so near V3. thats square E898 differs from the number 2 ony bythe small quantiy seo by which i exceeds it 788, We may proceed in the same manner. when itis required to find by approximation cube roots biquadrate roots, Be. Let there be given the equation of the third degree. 2° = a: or let it be proposed to find the value of ¢/@. ‘Knowing that itis nearly rm. we shall suppose x =n +p: neglecting p? and p', we shall have 2? = n-+ 3np so that 3n¢p = a~ n°, and p So whence oe f Sone q n+p) This the met given by Sirs, Newlon atthe Bezianing of his “Method of Foxons.” When investigate. itis found subject to diferem Imperfections: for whieh reason we may with advantage subsite the method given hy M. de a Grange, im the Memoirs of Ben fr 1968 aod Vere. ‘This metho has since been published by De la Grange. ina separate Tease, where the subee isSCUSE inthe usual mastery ile of his sucha.Inrodueing Undergraduates tothe History of Mathematics 25 If, therefore, n is nearly = /@, the quantity which we have now found will be much nearer it. But for still greater exactness, we may again substitute this new value for n. and so on. For example, let 2° = a = 2: and let it be required to determine 3/2. Here. if n is nearly the value of the number nt sought, the formula will express that number still more nearly: let us therefore make a and we shall have 2 2. n= 4.and we shall have 2 = 8, #2, and we shall have 2 3.0 cas 789. This method of approximation may be employed. with the same success. in finding the roots of all equations. To show this, suppose we have the general equation of the third degree, 2° + az? + br + c=, in which n is very nearly the value of one of the roots. Let us make x =n — p: and. since p will be a fraction, neglecting the powers of this Teter, which are higher than the first degree, we shall have 2? = n? — 2np, and 2? =n? — Sn?p; whence we have the equation n° — 3p + an* ~2anp + bn — bp+c=0. of n? + an? +i + c= 3n2p + 2anp~ bp = (3n? + 2an +b)p, Sant e ni tant+bnte) _ 2nd tan? —c ieee and za n- (Sten tse Bn? + Ban +b n+ dan>b_) ~ Gn? sans 8 than the fir, being substituted for n. will furnish a new value still’ more accurate. 790. In drt apply this operation tan example et 2+ 2s ~ 350 0. in which a = 2,6 2, and ¢= -50, FEE wit be a value suit nearer te truth ‘ot being far from the true one. we shall suppose 7’ = 3. which gives us x and if we were to substitute this new value instead of n, we should find another still more exact. 791, We shall give only the following example, for equations of higher dimensions than the third Let2? = 62 +10. oF 2* ~ 610 = 0, where we readily perceive that Lis to small and that 2 is 10 great. Now, bea value not far from the tne one, and we make = n+p. we shall have 2° = n° +5ntp: and, consequently, so that p = ‘This value, which is more exact If is supposed to be nearly the value of one of the roots, Now, the assumed value of = ite nb = 5ntp=Gn-+ Gp 10: or Sntp~ Gp = 6n~ 10— And pint Gn + 10— an? +10 Wherefore p = cand 2 (= n+p) = If we suppose n we shall have x value is altogether inapplicable. a circumstance which arises from the approximated value of m having been taken much too small, We shall therefore make n= 2, and shall thus obtain x = 2 = $2. a value which is much nearer the truth And if we were now to substitute for n. the fraction $2, we should obtain a still more exact value of the root =. ¢) Program Maple for Newton’ algorithm <5 2x40 expand (subs (x =x + 2, P)1): foptP) (31);