Let $\{X(t), t = \cdots -1, 0, 1, \cdots\}$ be a $P$ dimensional zero mean stationary Gaussian time series, $X(t) = \begin{pmatrix}X_1(t)\\X_2(t)\\\vdots\\X_P(t)\end{pmatrix}$ we let $R(\tau) = EX(t)X' (t + \tau)$, where $R(\tau) = \{R_{ij}(\tau), i,j = 1, 2, \cdots P\}$, and $F(\omega) = (2\pi)^{-1} \sum^\infty_{\tau=-\infty}e^{-i\omega\tau}R(\tau)$. It is assumed that $\sum^P_{i,j=1} \sum^\infty_{\tau=-\infty} |\tau| |R_{ij}(\tau)| < \infty$, and hence $F(\omega)$ exists and the elements possess bounded derivatives. It is further assumed that $F(\omega)$ is strictly positive definite, all $\omega$. Knowledge of $F(\omega)$ serves to specify the process. $F(\omega)$, and $S$, the covariance matrix of $x = \begin{pmatrix}x_1 \\ x_2\ \\ vdots\\x_P\end{pmatrix}$, a Normal $(0, S)$ random vector are known to enjoy many analogous properties. (See [7].) To cite two examples, the hypothesis that $X_i(s)$ is independent of $X_j(t)$ for $i \neq j = 1, 2, \cdots P$, any $s, t$, is equivalent to the hypothesis that $F(\omega)$ is diagonal, all $\omega$, while the hypothesis that $x_i$ is independent of $x_j$, for $i \neq j = 1, 2, \cdots P$ is equivalent to the hypothesis that $S$ is diagonal. The conditional expectation of $x_1$, given $x_2, \cdots x_P$ is \begin{equation*}E(x_1\mid x_2, \cdots x_P) = S_{12}S^{-1}_{22}\begin{pmatrix}x_2 \\ \vdots \\ x_P\end{pmatrix}, S = \bigg(\begin{array}{c|c} S_{11} & S_{12} \\ \hline S_{21} & S_{22}\end{array} \bigg)\end{equation*}. The corresponding regression problem for stationary Gaussian time series goes as follows. If \begin{equation*}E\{X_1(t)\mid X_2(s), \cdots X_P(s), s = \cdots -1, 0, 1, \cdots\} = \sum^P_{j=2} \sum^\infty_{s=-\infty} b_j(t - s)X_j(s)\end{equation*} then $B(\omega)$, defined by $B(\omega) = (B_2(\omega), \cdots B_P(\omega)), B_j(\omega) = \sum^\infty_{s=-\infty} b_j(s)e^{i\omega s}$ satisfies \begin{equation*}B(\omega) = F_{12}(\omega)F_{22}^{-1}(\omega), \quad F(\omega) = \bigg(\begin{array}{c|c}f_{11}(\omega) & F_{12}(\omega) \\ \hline F_{21}(\omega) & F_{22}(\omega)\end{array} \bigg).\end{equation*} It is interesting to ask how well these and similar analogies carry over to sampling theory and hypothesis testing. Goodman [3] gave a heuristic argument to support the conclusion that $\hat{F}_X(\omega_k)$, a suitably formed estimate of the spectral density matrix $F(\omega_k)$ has the complex Wishart distribution. The question is met here by the following results. Firstly if $\hat{F}_X(\omega_l), l = 1, 2, \cdots M$ are estimates of the spectral density matrix, each consisting of averages of $(2n + 1)$ periodograms based on a record of length $T$, with the $\omega_l$ equally spaced and $(2n + 1)M \leqq \frac{1}{2} T$, then it is possible to construct, on the same sample space as $X(t), M$ independent complex Wishart matrices $\hat{F}{\bar{\bar{X}}}(\omega_l), l = 1, 2, \cdots M$ such that $\{\hat{F}_X(\omega_l), l = 1, 2, \cdots M\}$ converge simultaneously in mean square to $\{\hat{F}_{\bar{\bar{X}}}(\omega_l), l = 1, 2,\cdots M\}$, as $n, M$ get large. Secondly, it is legitimate to use the natural analogies from multivariate analysis to test hypotheses about time series. One example is presented, as follows. The likelihood ratio test statistic for testing $S$ diagonal is $|\hat{S}|/\mathbf{\prod}^P_{i=1} \hat{s}_{ii}$ where $\hat{S} = \{\hat{s}_{ij}$ is the sample covariance matrix. The analogous statistic $\psi$ for testing $X_i(s), X_j(t)$ independent, $i,j 1 = 2, \cdots P$ from a record of length $T$ is $\psi = \prod^M_{l=1} \lbrack|\hat{F}_X(\omega_l)|/\prod^P_{i=1} \hat{f}_{ii}(\omega_l)\rbrack$ where $\hat{F}_X(\omega_l) = \{\hat{f}_{ij}(\omega_l)\}$ are the sample spectral density matrices as above. Letting ${\bar{dbar{\psi}}} = \prod^M_{l=1} \lbrack|\hat{F}_{\bar{\bar{x}}}(\omega_l)|/\prod^P_{i=1} \hat{h}_{ii}(\omega_l)\rbrack$ where $\hat{F}_{\bar{\bar{x}}}(\omega_l) = \{\hat{h}_{ij}(\omega_l)\}$ are the independent complex Wishart matrices referred to above, we show $EC_{n,M} |\log \psi - \log {\bar{\bar{\psi}}} \rightarrow 0$ for large $n, M$, where $C_{n,M}$ are chosen to make the result non-trival. The method of proof applies to any statistic which is a product over $l$ of sufficiently smooth functions of the entries of $\hat{F}_X(\omega_l)$. Applications to estimation and testing in the regression problem will appear elsewhere [8]. The distribution theory of functions of complex Wishart matrices has been well investigated by a number of authors [3] [5] [6], and hence can be easily applied here to statistics like ${\bar{\bar{\psi}}}$. The results above are shown for $P = 2$, it is clear that the proofs extended to any (fixed) finite $P$. The proofs proceed as follows, via a theorem which has somewhat more general application. For each $T$, let $X$ be the $2 \times T$ random matrix $X = \binom{X_1}{X_2} = \begin{pmatrix}X_1(1), \cdots, X_1(T)\\X_2(1), \cdots, X_2(T)\end{pmatrix}$ and let the $2T \times 2T$ covariance matrix $\Sigma$ be given by $\Sigma = \begin{pmatrix}\sum_{11} \sum_{12} \\ \sum_{21} \sum_{22}\end{pmatrix}$ where $\Sigma_{ij} = EX_i'X_j. \{\hat{F}_X(\omega_l)\}$, the sample spectral density matrices described above based on a record of length $T$, are each of the form $\hat{F}_X(\omega_l) = T^{-1}XQX'$ where $Q$ is a $T \times T$ circulant matrix with largest eigenvalue $ = T(2n + 1)^{-1} \leqq \frac{1}{2}M < <T$. We define circulant matrices $\bar{\Sigma}_{ij}$ which approximate $\Sigma_{ij}$, and a random matrix $\bar{X}$ on the sample space of $X$, $\bar{X} = \binom{\bar{X}_1}{\bar{X}_2} = \begin{pmatrix}\bar{X}_1(1), \cdots, \bar{X}_1(T)\\\bar{X}_2(1), \cdots, \bar{X}_2(T)\end{pmatrix}$ with $E\bar{X}_i'\bar{X}_j = \bar\Sigma_{ij}$. The $2T$ eigenvalues of the block circulant matrix $\bar\Sigma = \begin{pmatrix}\bar\Sigma_{11} \bar\Sigma_{12} \\ \bar\Sigma_{21} \bar\Sigma_{22}\end{pmatrix}$ will be the $2T$ eigenvalues of the $T$ matrices $\{F(2\pi j/T),j = 1, 2, \cdots T\}$. The distribution of random matrices of the form $T^{-1}\bar{X}Q\bar{X}'$ where $Q$ is any circulant matrix are relatively simple to investigate due to the fact that all circulant matrices commute, and their eigenvalues may be exhibited as simple functions of the elements. Circulant quadratic forms in random vectors with circulant covariance matrices are well known in the literature, (See [1] and references cited there). Let $\hat{F}_{X,Q} = T^{-1}XQX'$ and $\hat{F}_{\bar{X},Q} = T^{-1}\bar{X}Q\bar{X}'$ where $Q$ is now any $T \times T$ (real or complex) quadratic form with largest absolute eigenvalue $\leqq q$. The main Theorem allows the replacement of $X$ by $\bar{X}$ in the analysis, and is, that under the assumptions on $F(\omega)$ and $R(\tau)$, for any $T$, \begin{equation*}\tag{1.1} E \operatorname{tr} (\hat{F}_{X,Q} - \hat{F}_{\bar{X},Q})(\hat{F}_{X,Q} - \hat{F}_{\bar{X}, Q})^{\ast'} \leqq cq^2/T^2\end{equation*} where $c$ is a constant depending only on $F(\omega)$ and $R(\tau)$. A lemma, essentially allowing the replacement of $F(\omega)$ by a suitably chosen step-function, together with the application of (1.1) gives the results concerning the $\{\hat{F}_X(\omega_l)\}$ an $\lambda$. Since $\hat{R}(\tau)$, the sample (circularized) autocorrelation function is also of the form $T^{-1}XQX'$ with $Q$ circulant we obtain an easy corollary on the distribution of $\{\hat{R}(\tau)\}$.
A gas or vapor bubble moving in translation in a liquid and varying its volume propels itself by a rocket effect, which occurs in foams moving in pressure or temperature fields with a non-zero gradient. It is notably observed in boiling and centrifugal pump operation, which latter features the following two characteristic bubble types : __ 1. Gas bubbles, which are apt to emerge faster from the pump than the liquid; 2. Vapor bubbles (cavitalion). When a bubble in translatory motion implodes, energy is transferred: potential pressure energy is converted info kinetic implosion energy, which in turn becomes kinetic translation energy. The instrument for this energy transfer process is a "micro-jet" following the bubble. It is shown that a considerable increase in kinetic translation energy density occurs, which is converted into potential pressure energy on impact against a solid obstacle. The resulting pressures (10,000 kg/sq.cm) explain the mechanical aspect of cavitation erosion. Several experimental results have confirmed these theoretical considerations. The development of a rotoscope is also described, this being a device enabling a centrifugal pump impeller (for example) to be "stopped" so that only the relative motion of the flow particles remains. Unlike with a stroboscope, time exposures can he taken with this device. With more thorough knowledge of foam mechanisms, and especially of cavitalion bubble behavior, it should be possible to design cavitation erosion-proof impeller blades.
This paper is concerned with repetitive sequential play in finite statistical games (decision problems) from the statistician's point of view. We shall assume that the statistician's move at stage $k$ may depend on the previous $k - 1$ moves of Nature as well as the random variable $\mathbf{X}_k = (X_1, \cdots, X_k)$, where the $X_i$ are independent observations (r.v.'s) (possibly vector-valued) from the sequence of statistical games, $k = 1, 2, \cdots$. The play is repetitive in the sense that each component game is identical in structure, with only the moves of the statistician and Nature changing. Furthermore, we impose no assumptions regarding the behavior of the parameter sequence of Nature's moves. The statistician does have the added disadvantage that the finite class of distributions in the component game is not fully specified. However, he does know that class in question has: either (i) all members with discrete distributions or (ii) all members with $q$-dimensional a.e. continuous Lebesgue densities. This same problem when the distributions are fully known has been treated in [6] for statistical as well as more general games in which Nature's space is finite. In the case where the distributions are completely specified but the history of the past moves is unknown to the statistician, see [20], [22], [27], and [28]. The development in this paper is closely connected to and motivated by these results, particularly those of the preceding paper [27]. If for fixed $N$, the empirical distribution $p_N$ of Nature's moves is known, then the statistician could use as a rule for each of the $N$ component games a strategy Bayes against $p_N$ having risk $\phi(p_N)$. In all the papers cited in the previous paragraph, the aim was to construct for the statistician, when $p_N$ is unknown and $N$ not specified, a sequence of randomized decision functions whose $N$th average loss minus $\phi(p_N)$ approaches zero (or has an upper bound approaching zero) in a suitable sense as the number of repetitions of play, $N$, increases. However, in the case of statistical games, all of the above results require that the finite class of distributions be fully specified. In this paper we remove that assumption by estimating the distributions sequentially based on past moves and observations. Then in the present play of the component game the statistician substitutes these estimators into a procedure which is Bayes against the empirical distribution of Nature's previous moves. The resulting sequence of procedures is shown to be "asymptotically good" in the sense that the average loss over the $N$ games $W_N$ minus the Bayes risk $\phi(p_N)$ approaches zero (in an appropriate sense) as $N$, the number of games played, increases. In Section 2 we introduce notation and preliminaries. Section 3 discusses play in repetitive games and defines the proposed sequential procedures $\mathbf{t} = \{\mathbf{t}_k\}$. In Section 4 we prove preliminary results upon which all proofs are founded. Section 5 considers the discrete case giving uniform (in sequences of Nature's moves) convergence theorems (as $N \rightarrow \infty$) for the quantity $W_N - \phi(p_N)$. Theorem 5.1 is a uniform convergence theorem of $O(N^{-\frac{1}{2}})$ of the expected value of $W_N - \phi(p_N)$ for finite discrete classes, each member of which is non-degenerate and satisfies a certain tail probability condition. Under the same conditions, Theorem 5.2 gives uniform convergence to zero in probability for the quantity $N^{\frac{1}{2}} (\log N)^{-1} \{W_N - \phi(p_N)\} \text{as} N \rightarrow \infty$. Uniform convergence of $W_N - \phi(p_N) \rightarrow 0$ in probability for general non-degenerate finite discrete class is presented in Theorem 5.3. Section 6 treats the estimation problem for densities needed to form the randomized strategy sequences $\mathbf{t}$ in the continuous case. The results stated are based on a paper by Cacoullos [3] generalizing the univariate results of Parzen [15]. In Section 7, we present results for the continuous case. Theorem 7.1 and its corollary give uniform convergence of $W_N - \phi(p_N)$ to zero in probability and of its expectation to zero, respectively. The finite continuous classes of Theorem 7.1 are very general in the sense that each member is a continuous a.e. density. Finally, in Section 8 we draw certain conclusions and relate our results to similar results obtained elsewhere. The novelty of the paper rests in the fact that through the past history of Nature's moves and the observations connected with past play, one can construct a sequential strategy, $\mathbf{t} = \{\mathbf{t}_k\}$, with very little knowledge about the finite class of distributions, which approaches asymptotic "optimal" play. The lack of knowledge on the finite class of distributions distinguishes this work from the related "repetitive type" problems in games and/or decision theory treated in [1], [2], [4], [6], [7], [8], [9], [10], [12], [17], [18], [19], [20], [21], [22], [24], [25], [26], [27], [28], and [29]. For possible applications of this work see Neyman [14], especially his Example 3 and his discussion relating to the work of Blackwell [2].
Professor Keeping's book is a text for a one-year course (90-100 hours) for students having a knowledge of elementary calculus-second or third year students.It is unusually complete in that it is difficult to think of a topic which is not treated, at least briefly, but which one might like to see included in such a course.As would be expected of a widely ranging book at this level, many results are stated without proof but it is by no means a " how to do it" book.In addition to the usual elementary probability theory, standard distributions, and classical estimation and testing, one finds, e.g. the cumulants and Ar-statistics, sampling techniques, sequential and nonparametric procedures, fixed, random and mixed models as well as latin square and incomplete block designs considered, and a last chapter which looks at multivariate problems and introduces stochastic processes.There is a laudable concern for the power of the tests discussed and the required non-central distributions are introduced.Appropriate tables, a large number of exercises (with answers) and a thirty page appendix on various mathematical topics are included as well.The price paid for the virtue of comprehensiveness is, of course, the brevity of some particular parts; one cannot have everything.However, one might reasonably suggest that the briefer the treatment the more precise should be the statements.This book is somewhat marred by puzzling, misleading, or false statements, e.g. both the sample and population moments are defined to be the " rth moment of X about zero"; "If T is sufficient, so is any function of T" (p.125); the variance of a maximum likelihood estimator is asserted to be the Cramer-Rao lower bound; in discussing the Mann-Whitney U-test, it is not clear at given points just what alternatives are being considered and while one statistic is described as the test statistic, we are instructed to reject for small values of another.
The nonrelativistic potential energy between two spinless particles is deduced from a knowledge of all phase shifts at a given energy. A spherically symmetric potential is found always to exist, but it is not unique. In particular, for every energy, there exists at least one nonzero potential which causes the scattering cross section to be zero. The paper contains both the formal construction procedure and the necessary existence and uniqueness (or lack of it) proofs. Some general examples are included.
One of the features of the text is an elaborate code which is used to refer to certain axioms, definitions, and theorems.For example, TIr is the Theorem on Irrational Numbers, which runs as follows: "If a non-zero rational number 'r is combined with an irrational number p by any one of the four operations of arithmetic, the result produced is an irrational number; in symbols, r + p, r -p, p -r, rp, r/p, p/r are irrational numbers."According to the author 1 s preface, "Experience in classroom teaching shows that the students use the code with alacrity and effectiveness in making full and concise proofs, " This reviewer feels that the book under review is a worthy addition to the literature; but on the whole he found the exposition somewhat clumsy.In a few places terms are used before they are explained (e.g."empty set," page 99) and in c some places no explanation is offered where one is clearly required, (e.g.01 is used, but never defined.Since 31 is defined, the reviewer presumes that no knowledge of factorials is assumed.)Functions are never mentioned, even though the use of functions could have simplified the treatment considerably.These objections, however, may possibly be regarded as minor.Finally, the exercises in the book are many in number and generally non-computational in nature.
exists for every i (Theorem 9 of [3]).Following Levy the state i is called stable or instantaneous according as gt-as finite or infinite.We refer to [2 ] for the foundations of the theory of Markov chains under consideration.Although knowledge of these foundations will be necessary for a thorough understanding of what follows, we shall strive to make the present paper readable by itself.Let i be a stable state with q,>0; such a state always exists unless P{x(t) =x(0), 0^t<°° }=l [8, p. 375].Suppose that P{x(0)>=*} =1.LetX=X,(w) be the "first sojourn time" in the state i, namely the length of the first tinterval in which x(t, w)=i (seeTheorem 1 of [2]).Then P{\^t} =l-e~">', t^O [3, p. 54].Let j be an arbitrary state (not oo!) and defineIf J9*i, a is the "first entrance time intoj"; ii j = i, a is the "second entrance time into i" (the first being zero by hypothesis).It is easily shown that a is a random variable in the broad sense, namely a measurable w-iunction defined on a measurable w-set whose probability may be less than one.We define its distribution function in the broad sense by Fij(t)=P{a^t}.Now it can be proved that the two random variables X and a-X (which may be zero with positive probability) are independent^).This is a special case of Theorem 5 of [2], but we give a simple proof as follows.Let us first note that the distribution of a -X may be derived as follows.It can be shown that(3) a(w) considered as a point on the i-axis is the limit from the right of points of Sj(w) ={t: x(t, w) =j}; hence we havewhere h = 2~m, ra->• oo.Now for each 5 > 0 define two random variables X,=X,(w) and a, = aa(w) on the set {w: x(s, w)=i} as follows: X"(w) is the supremum of T such that x(t, w)=i, s^t<s + T; as(w) is the infimum of t such that t>\"(w) and x(t, w) =j.Thus X0 and a0 reduce to the previous X (2) The random variables Zi(w), ■ ■ ■ , z"(w), with domains of definition Ai, • ■ • , A", are said to be independent iff P{ flLx A*[z*M Set] }/P{ f|Li A*l = LTt-i P{A*tz*W £c*]}/P(At) for every real Cx, • • • , ck.(3) In fact, the set Sj(w) is dense in itself (see [2, §4, (ii)]).
The applicability to a quantum liquid of the standard classical formula connecting the compressibility with the coherent scattering cross section for large wavelengths, questioned by the author in a previous paper, is examined. The correctness of the standard formula is proved (a) at absolute zero (the density fluctuations being infranormal); (b) under quantum conditions for all temperatures at which the Wigner expansion converges (it is conjectured that for liquid helium the expansion may diverge below the lambdapoint); and (c) for a one-dimensional crystal for all temperatures. These results, while they stop short of a complete proof of the standard classical formula for all conditions, do extend considerably our knowledge of its range of validity.
The possible methods of utilizing magnetrons to generate short-wave oscillations are indicated and the more important results of previous workers are described. The theoretical basis of “electronic”and “dynatron”oscillations is discussed, with particular reference to those features which can be investigated experimentally. It is shown that the wavelength of the electronic oscillations is determined mainly, if not entirely, by the electron time of transit. The general expression for the time of transit is given and hence expressions for the wavelength in terms of magnetic field strength are obtained for zero and saturated space-charge conditions. The wavelength is found to be about 36 per cent greater in the latter case. It is shown that initial electron velocity causes an appreciable reduction in wavelength in normal cases. The effect of magnetic field on space charge is considered and is found to lead to a small, possibly negligible, increase in wavelength. It is shown that the space charge is uniformly distributed when the magnetic field strength exceeds the critical value at which the electron orbits just touch the anode surface. This result has been previously stated by Hull, but Hull's deduction that the electrons travel in circular orbits round the cathode is disputed.An attempt to provide a simplified theoretical explanation of the “dynatron”characteristics of a “split anode”magnetron leads to a false result from which it is concluded that no theory will provide an explanation which does not take into account the non-uniformity of the electric fields in the two halves of the valve. The general shape of the static characteristics is indicated by means of Habann's theory, which is, however, not capable of giving proof of the existence of negative resistance in the case of a symmetrical oscillatory circuit, which is the case considered here. A qualitative explanation of the occurrence of negative resistance, i.e. of the greater fraction of the anode current reaching the lower-potential anode segment, is given.The object of the experimental investigation was to discover the nature of the fundamental relations in the electronic and dynatron types of oscillation, to compare these relations with the indications of the theory, and to apply the knowledge obtained to the production of a sufficient amount of power to be technically useful at the shortest possible wavelength.For electronic oscillations it is found that the experimental results are entirely in agreement with the theory in so far as it is applicable. In particular it is confirmed that the strength of the electronic oscillations is greatest at the “critical” relation between anode voltage and magnetic field strength, and that the wavelength of the optimum oscillation is inversely proportional to the magnetic field strength. The actual value of the wavelength and the amount of the wavelength change due to space charge both agree with the theoretical values within experimental accuracy. It is concluded that the effect of magnetic field on space-charge distribution is not great enough to affect the wavelength appreciably. It is shown that apparently anomalous results can be explained by taking into account the stray capacitances, due in particular to the glasswork of the valve, across the oscillatory circuit. Theselead to internal resonance effects which, although often a nuisance to the investigator, sometimes have the advantage of enabling a relatively large output to be obtained at a particular wavelength.The fact that the greatest output is, in general, obtained with the magnetic field not exactly in the direction of the electrode axis, which has been reported by Slutzkin and Steinberg and by Ranzi, was observed independently. The existence of an optimum field angle differing from zero is found to be due to the resultant spiral motion of the electrons balancing out the effect of cathode potential-drop for electrons arriving at part of the anode surface in such a way as to increase the number of oscillating electrons.By making use of an internal resonance effect and suitably adjusting the field angle,'an output of the order of 1–5 watts was obtained at a wavelength of about 24 cm.It is pointed out that the wavelength of the electronic oscillations can be expressed in the same form as the Bark-hausen-Kurz equation for the triode case, and that the existence of an optimum value of anode current (approximately 1/10th of the space-charge saturation value) leads, as in the triode case, to the result that the minimum wavelength obtainable without overloading the valve depends on the anode diameter and decreases with it. For an anode diameter of 3 mm the shortest wavelength at which optimum oscillating conditions can be maintained is of the order of 20 cm. The corresponding figure for the triode case (grid diameter 3 mm) is about 50 cm. The shortest wavelength actually observed was 18 cm. Oscillations of shorter wavelength have not been investigated, owing to the small power obtainable.It is shown that it is possible to obtain dynatron oscillations in a split-anode magnetron by tilting the magnetic field, and that these are distinct from both the electronic and simple dynatron oscillations. Hollmann has investigated the occurrence of oscillations of this kind in the full cylindrical, anode magnetron and found a minimum wavelength of about 20 m. It is shown here that these oscillations can be produced down to about 35 cm wavelength, but at this wavelength the effect of electron inertia is important.In the investigation of the simple dynatron oscillations, static characteristics showing the negative resistance effect are obtained. From these curves the operating characteristics of the valve are calculated and checked by comparison with a. set of measured values (for a relatively low frequency). Good agreement is obtained.It is found that the energy of the dynatron oscillation starts to fall off rapidly at a wavelength which is about 4 times the electronic oscillation wavelength corresponding to the anode voltage used. This leads to a formula for the wavelength limit for dynatron oscillations.The relation between oscillation amplitude and anode voltage is discussed, with particular reference to modulation.It is found that during oscillation the anode current may exceed the original total emission. This is probably due to bombardment of the filament by electrons which return to it with considerable velocity. The exact mechanism of this bombardment is not clear and the effect is being further investigated. By taking steps to reduce this effect it has been possible to obtain an output of 60 watts from a relatively small valve at about 2 m wavelength.The shortest wavelength obtained by means of dynatron oscillations was about 30 cm. At this wavelength the power obtainable was about 0-1 watt. It is concluded that for wavelengths below about 50 cm electronic oscillations give the greater output.
Part I. Introduction 1. Prologue.Thus far, very little has been published on the general theory of formal modular invariants or covariants.Workers have, on the whole, obtained results for special, more or less isolated, cases; and although some beautiful and important general theorems have been proved, they are more or less unrelated.This is, of course, only natural in any division of knowledge in its formative state.Nevertheless, no worker in the field could fail to be conscious of a certain uniformity common to the special cases that have been studied in detail; though (alas !) this uniformity usually appeared to be broken ruthlessly in the next case studied.This breaking of an apparent law signified, however, merely that we did not know these special cases with a sufficient thoroughness of illuminating detail, or were trying unwittingly to make the laws conform to certain standards, unconsciously preconceived.This latter handicap was laid on us naturally enough by our thorough knowledge of algebraic invariants and the fact that this newer kind of covariants is, in many ways, strikingly like the older, classic covariants, though so tantalisingly different.Their similarity and their difference show themselves in the very beginning of the study: in the definitions, in the simplest examples.Perhaps the differences that first come to mind are those which are inherent in the fields of definition, which, in the case of classic covariants, is the field of reals or ordinary complex numbers and, in the case of modular covariants, is a Galois field, GF[pn], of order pn.These differences are too obvious to mention in detail, but one who has studied the beautiful proofs given by the old masters of invariant theory has been forced to the conclusion that most of the proofs seemed to use the properties of a field of characteristic zero, not in some accidental manner, but rather in veriest necessity.Growing from the surface differences between the two fields are two very important distinguishing characteristics of the two kinds of covariants.It * Part II was presented to the Society, September 7, 1920; Part III, December 28, 1921; Parts IV and V, December 27, 1922.