BE561614A - - Google Patents

Description

       

    <Desc / Clms Page number 1>
 



   The object of the present invention is to provide a new composition system for printing texts, generally aimed at accelerating this composition, at least at making it largely automatic, and therefore at reducing the costs of printing. said composition.



   It also aims to establish this new system such that it makes use of the current possibilities of automation of operations, even of an intellectual nature, and that it therefore reduces human intervention in this composition to the small number. only intelligent actions concerning the understanding of texts and their aesthetic presentation.

  <Desc / Clms Page number 2>

 



   Its further and additional object is to provide for this new system such that it involves the automation of the cutting and justification operations, customary in east composition, which until now required the intervention of specialized operators.



   The current process of preparing and composing a text can, it seems, be summed up as follows:
An author's manuscript, possibly modified, gives rise to the establishment of multiple typewritten copies.



  This is a cleanup operation.



   On one of these typed copies, then works a composition preparer. This work, in addition to the obvious corrections of typing and syntax errors which may have remained despite the corrections previously made to the copies, consists essentially in the transfer to the said copy of the following indications or service information: a. Indication of the format and possibly the margins; b. Determination of justification; vs. Determination of the fonts, varieties and body of characters of letters, numbers and signs, in general, of the elements of the text to be composed; d. Possible indications of modifications of paragraphs in the typewritten copy; e. Indication of reservations for printing of clichés; f.

   Possible indications of the locations of footnotes in the composition, etc.



   The copy thus prepared is then delivered to a printer who, having access to the text and its operating instructions, must ensure the control of a composition machine, while also ensuring two distinct and essential operations: the cut at the end. line and justification. It doesn't matter

  <Desc / Clms Page number 3>

 moreover, from the point of view of the present invention, that the immediate result of the intervention of an operator at the printer's resides in the final production of a lead-coated alloy composition or of a photographic film impressed, or simply in the printing (most often by perforating a ribbon) of a support which will then be used to control a semi-automatic or automatic composition machine.

   It should only be recalled that such automatic or semi-automatic control composing machines are known, but that in such a case, until now, the preparation of the control tape had to be ensured by the intervention of a specialized operator.



   A more precise object of the present invention is above all to provide a new method, and a new system implementing this method, which eliminates in the composition of a text to be printed any intervention at the printer of a specialized operator to decide on the cutting and justification operations of composition lines.



   It also has, complementarily but not necessarily as a particular object to provide for this process and this system such that they make it possible to avoid any operation of clearing a manuscript, prior to the composition, for burping however that this manuscript is presented. to the composition preparer in the form of a legible copy, preferably typed.



   In what follows, "service information !!" denotes all information and indications of the kind that have been defined above for the work of preparing the composition and the composition itself.



   The term "coded record" is reserved for any record of information liable to be subject to a

  <Desc / Clms Page number 4>

 mechanical, electrical and (or) optical probing to provide electrical signals whose specific configurations simulate the information elements of these recordings.



  The term "unencrypted recording" is reserved for any recording of information capable of being read directly, even if the signs it contains are, in part at least of an arbitrary nature, of the stenographic or shorthand type. typical for example.



   The composition process in accordance with the invention is then essentially characterized in that it comprises the following operating stages, both manual and automatic: a) Establishment of at least one coded record of the information elements of a text and elementary service information accompanies them - this operational stage can include a period or phase of establishment of this or these recordings which is unique or which can include several indifferently, and which can give rise both to the establishment of a mixed recording, text and service, as distinct recordings of text on the one hand, of service on the other;

   b) Progressive conversion of this or these coded recordings into as many sets of electric signals - this second stage possibly also comprising either a single phase, or as many alternating phases as necessary, without distinction; c) Transposition of these sets of electrical signals into at least one set of signals, or several if necessary, then defining or describing the complete typographical composition of the text as it should be carried out by a machine. dialing, manual or automatic;

   

  <Desc / Clms Page number 5>

 d) Establishment of recordings of the set or sets of signals delivered as a result of operative step c, whether these latter recordings are coded with a view to semi-automatic or automatic control of a mechanical composition machine, or are in clear with a view to manual control of such a machine.



   This process is further characterized by the following main variations which can be introduced in its developmental stages:
1) During stage a, and when necessary, established sow at least one coded record of correction of the text and / or related service information - during stage b then, conversion of this coded record from corrections to a set of electrical signals for cancellation and substitution for signals from the first recording, in view of stage c;
2) During stage a, making at least as many unencrypted recordings as there are coded recordings;

   
3) During the stage of simultaneous establishment of at least one coded recording and one clear recording - these variants can be separately or simultaneously introduced in the above-defined process.



   An apparatus for the implementation of this method is then characterized, and generally, in that it comprises, in operative combination, at least means for establishing the above-defined coded recordings, means for translate these coded recordings into electrical signals means for transposing these signals into at least one set of signals describing the complete typographical composition of the

  <Desc / Clms Page number 6>

 text, and means for recording these latter signals.



   The practical use of the invention, now, could of course resort to a unitary realization, more exactly localized, but then the overall rate of operation would be just as clearly conditioned by the development of the coded recording of service information and this information requires the intervention of a human factor.

   Consequently, and in order to ensure, in the practical use of the invention, full use of the qualities of speed and flexibility intrinsic to said method and to said system, the invention provides for resorting to rational fractionation of the means of implementation, namely:
On the one hand, the means specific to the establishment of the various coded records defined in the operational stage a;
On the other hand, the means specific to analyzing the information from these recordings, to their electrical conversion and transposition and to the recording of the information thus transposed.



     Seen from this last aspect, a composition system for implementing the method according to the invention is characterized more particularly in that it comprises at least one apparatus for producing coded records, and preferably a plurality of coded recordings. these devices, and at least one device fed by the recordings coming from the first one and established to form, from the automatic examination of these recordings, the record or records of complete typographic composition of the text for control of a composition machine.

  <Desc / Clms Page number 7>

 



   Any device for the preparation of text records, service information, and corrections, can advantageously comprise a printer machine with manual keyboard control, according to the indications which will be specified later.



   Any apparatus for producing the record (s) ultimately desired may advantageously include an electric or electronic calculating machine, of the type of program calculators. This program will also be specified later. It is, however, well known that such calculators are directly suitable for entering data in the form of tapes of coded recordings and for delivering results also in the form of recorded tapes, coded or in clear.

   In addition, any calculator of the so-called universal type can, without modification of its internal organs and by the only modification of its program memory, see its operation adapted to the automatic resolution of all problems concerning the data entered, and consequently to any problem. entering into the category of operations of cutting, justification and indication of the typographical composition of a text, the latter being essentially based on counts of signs and blanks and comparisons to predetermined references.



     Illustratively, and with a view to clearly specifying the characteristics, methods of implementation and advantages of the invention, various examples of systems in accordance with the above-defined process will be detailed with reference, when necessary, to the accompanying figures. , whose subjects will be specified as this description proceeds.



   We will first consider various methods of ela-

  <Desc / Clms Page number 8>

 boration of coded recordings to be processed later for transposition.



   In the event that the publisher has only a manuscript which is unsuitable for the direct establishment of a coded text recording, and therefore together with a coded service information recording, the clarification of this manuscript must be performed. It is then possible to proceed, simultaneously with this clarification, to the establishment of the coded recording of the text and also to the establishment of the coded recording of the service information, preferably.



   Two cases may arise: one where the composition preparer can work on the manuscript before it is edited; the one where the focus must precede the work of the composition preparer.



   In the first case, it seems more advantageous to carry out simultaneous sharpening of the text and of the service information added to this text by the preparer. The "typing" of the machine printing the text and the information in net, e @ service will ensure the simultaneous production, in addition to a recording in clear on page or double page, of a coded recording tape, without that the typing operator does not have to worry about the production of this tape by the machine other than by observing instructions such as the following: do not go backwards in a line as well as from one line to another; type in the margin of each line its number in the copied text (if the machine is not automatically prepared for this typing on return of the carriage);

   note your own copying errors (optional) but do not correct (imperative).

  <Desc / Clms Page number 9>

 



   The coded tape - the term "tape" being taken to designate any kind of recording medium and any kind of printing on this medium - can be produced by the machine either with two tracks, one for the text, the another for the coded information, or again and preferably with a single track comprising the two recordings with their elements of information interlaced.



   The operation will be resumed with a text and the related service information, for the establishment of a tape of corrections. For this ribbon to be unique, it will suffice for all the corrections to be made to the text to be brought together, whether the corrections come from the typist, from the composition preparer (or from the one who rereads the text put in the text. net) and (or) of the author who will have revised the said editing, etc. In this case, however, the typing operator will be required to enter the numbers of the corrected lines since these lines do not follow one another in normal numerical order.

   Any correction in a line will of course have to be made by retyping the entire line and, if a line is replaced by several, they will bear the same number, corresponding to that of the corrected line.



   In the second case, where the manuscript is unsuitable for the work of preparing the composition itself, it will first be cleared. A first encoded text recording tape can, if desired, be simultaneously produced under the same conditions as provided above. Then, the preparer having worked on a copy of this setting in net, the only coded recording of the information of service could be carried out, by a direct key or not, either on

  <Desc / Clms Page number 10>

 another support, or on the same support scrolling again under an inscription mechanism identical to the first - in this case, the tape will have two parallel tracks, as in a presentation variant already indicated above.



   The correction tape can always be done in the way specified for the first case.



   The reason why it would seem preferable to have only a single recording tape, single or dual track, is that the timing of the advance of the two information groups, text and service, will be. automatically ensured for their introduction into a calculator, without requiring special forecasts concerning the synchronism of several ribbons having to scroll in constant relative phase, for this data entry.



   By preparing, on the contrary, the data on two ru- bans only, one of first keystroke, the other of corrections, only remains in play, at the input of the calculator, the simpler problem of stopping one ribbon to drive the second and vice versa.



   The case where the first focusing tape and the correction tape are read independently for the input of data into the calculator, the latter then ensuring, by an internal subroutine, the comparison of these two data series and the substitution of the data of the second series with the corresponding ones of the first series.



  This operating mode would be limited in practice only by considerations of an economic nature, because it would obviously require memories with very large information capacities.

  <Desc / Clms Page number 11>

 



   It is also not excluded from the field of the invention that the preparation of the composition, on the one hand, and the corrections on the other, are carried out prior to any typing operation delivering a tape of encoded recordings, interlaced. or side to side. Only typing errors could remain, which would require re-reading line by line as the typing progresses and the cancellation of any line that contains a typing error, its replacement by @ following line being direct when the recording is provided at the input of the calculator.

   This would make any indication of the line number in the typing unnecessary.
Whatever solution is adopted, it is obvious that its choice is dictated by considerations of orders above all external to the proper development of the process and system in accordance with the invention: organization and economy of the work of preparation and of preparation of records.



   The printing machine forming the ribbons of coded records, at the same time as providing a keystroke of text and service information, may be derived from a typewriter or from a teletype machine. , especially in the case / where the tapes are separate, comprising only one track, or comprising double tracks. So in fact, it suffices to complete the arrangement of the typewriter by a printing unit on ribbon having its elements (electromagnets for example) actuated from the keys of the standard keyboard or completed for the purposes required for a preparation. decomposition.

   On the other hand, it suffices to omit, in a teletype device, the outputs for online sending of the electrical signals formed, while keeping the device.

  <Desc / Clms Page number 12>

 standard visual control of the teletype and its arrangement of coded ribbon perforation, In both machines, the printing of the line numbers can be ensured automatically by actuation of a step-by-step at each return of carriage, step by step controlling this printing on both the ribbon and the page. For the establishment of service records, an arbitrary correspondence can be established between the keys of the standard keyboards of these machines and the designations of said information.

   A tape or a track of text would automatically be interpreted as such by a calculator depending on the input on which it would be read; a tape or a service track would similarly be interpreted as such by the calculator.



   In most cases, however, the keyboard will have to be modified, especially in the direction of its enlargement of various characters and signs. In the case, in particular, where the text and service information is to appear interspersed on a single track, intermediate indications will become necessary so that the same numerical codes are then correctly interpreted by the calculator.



  Such indications will be called "announcement" in the remainder of the description, which will more precisely consider the case of such a tape in more detail, for the fact that it appears to be one which requires more adaptation. pushing of usual devices and machines with a view to its use. However, it must of course remain that the foregoing arrangements find their natural places in the field of the invention.



   The example illustrated in fig, 1 shows a fragment of tape of coded recordings in which alternate

  <Desc / Clms Page number 13>

 galley text elements and service elements.



  This ribbon was therefore prepared by typing all of this information. The ribbon portion of FIG. 1 can equally concern a typing of text and service or a typing of corrections, text and service.



   In this example, is used, without limitation moreover, a binary digital code with seven code elements. On tape 1, provided. a row of drive perforations 2, here central but being able to be laterally offset as desired, or even doubled, run seven lines designated by the references j, n, v, t, b, a, f for the seven elements of code, plus a line, called d, the utility of which will be specified later in relation to the example calculator which will be described.



   The presence of drive perforations presupposes a step-by-step advance, both for writing and reading. This is not imperative, although it is considered more advantageous from the point of view of precision and convenience of use, and a continuous unwinding with friction drive could just as well be provided.



   Whatever the training mode adopted, the inscription defines on the ribbon columns spaced by an interval marked @, predetermined, which, once converted into a time interval by reading, will define the duration of a The calculator's code moment, with all eight lines being read in parallel, just as they are printed in parallel when writing.



   This impression is taken here perforated. This is also not limiting and any other well-known printing method by inking, burning, electrolysis, recording

  <Desc / Clms Page number 14>

 magnetic, photography, etc., can be used. It is only a question of varying the nature of the support in certain cases and of substituting for the @ 'heads. @scription and reading for perforations the heads of constitution appropriate to the nature of the desired impression. Both are moreover well known in the art.



   Advancing from right to left, the ribbon of fig. 1 should read from left to right. He showed a beginning of tape, end of primer cut, from a moment when a justification information had to be given. The justification is expressed by a numerical value defining the length of the line of the text to be printed finally.



   In the form shown, the annunciation indications used are as follows: J: announcement of the justification code; announcement of the typed text line number code (using this generic typing term for all types of typing);
V: announcement of the numerical code defining the variety of signs, both the font and size of the characters and the groups that can be defined for the characters in the different sizes and fonts;
T: announces the numerical codes of characters in their own alphabetic or symbolic meanings;
B: announcement of a blank code or space between words in the typed text;

   end of typed line indication will also be given as a blank, unless the typed line ends with a hyphen to break a word, when typing, from one line to the next, in which case the blank end of line will be deleted when typed

  <Desc / Clms Page number 15>

 even.



   In addition. and insi that it is specified on the diagrams of fig. 2 and 3, showing portions of tape for these purposes, also an A start announcement indication and an F paragraph end announcement indication.



   If only a five-element code were to be used, the information of blanks and the beginning and end of paragraph could consist of specific numerical codes, preceded by the announcement T, for example, or the beginning and end of paragraph, by codes preceded by the announcement V, a blank being represented by a particular code preceded. of the T.



   On the ribbon of fig. 1, there is in the ad column J a single perforation in the line j. Two columns follow with seven element numeric code perforations plus a perforation in row d. The first column which follows the announcement column J makes it possible to describe in binary numeration a value between 0 and 127, the second, together with the first, allows eori re any value between 128 and 16383 inclusive.



   The line number of the typed text is announced by the presence of a single column perforation on line n; this is the announcement column N. Then, come two columns reserved for indicating the numerical value of the typed line of text. These two columns do not have any perforation on line d, but the codes used require the presence at least two perforations in other lines, to differentiate them when reading the ad columns.

   The omission of perforation in line d is dictated in the example, pending description; over there

  <Desc / Clms Page number 16>

 need to remove @@ eros of typed lines when entering data to arithmetic operator and calculator information memory. The numbering of lines of typed text can be repeated periodically, for example every ten thousand lines, without inconvenience.



   Next is a V ad column, featuring a single perforation on the v line of the tape. The following column, which has a perforation on line d, specifies by its other perforations the numerical value assigned to the information V, in binary numbering from 0 to 127.



   The announcement column T which then comes has only one perforation on the line t of the tape. It is followed for example by a column indicating the numerical value of the code of a letter of the text, plus a perforation in d:
One can assume illustratively that a blank should immediately come after this opening letter of the text. This blank (of the typewritten copy) is announced in column B by a single perforation on row b of the tape. In the next column, only one perforation is found on row d of the ribbon.



   After this blank, service information V and letters or signs of the typed copy may alternate. The portion of tape shown is cut to abridge the representation and arrive. at a typed end of line 'It can end with a blank end of line, on which comes the number of the next line of text. It could also end with a letter or a sign, in the event of a word break at the end of the line in the typed text.

   An end-of-line blank can be directly im-

  <Desc / Clms Page number 17>

 pressed on the ribbon by a carriage return of the machine used where it can be struck normally If the carriage circumference comes into play and the line struck ends in a dash, the strike of this one will put out 1 action the carriage return for automatic printing of end-of-line blank.



   For a beginning of paragraph, 'freeze 2, after the number of Ua line which must, thus begin in the typed text, comes a column A announcing the beginning of paragraph, bearing \ A perforation in the line d of the tape . The following information is as in fig, 1.



   For the end of the paragraph, fig. 3, after information as specified in relation to figure 1, comes an announcement column at the end of paragraph F, presenting a single perforation on line f of the tape, followed by a column not bearing than the only perforation in line d of the tape.



  Immediately follows the ad column N of the number of the next line of typed text. If the printing of a typed line end blank is provided for automatic, @@ pressing the end of paragraph end key will disable carriage return.



   From the composition of the strip which has just been described, any formation of separate strips or at least with separate tracks for the text and service information can be directly deduced. Of course, the information relating to the numbering of the lines of the typed text will be transferred to both of the separate ribbons.



   Alternatively, on a medium and a single track on that medium, the text information may alternate

  <Desc / Clms Page number 18>

 and the service information .., line by line of the typed text, the numbering information of the lines of this text being for example spelled out for the information of a line of text then for the information of a service line, in order to allow the calculator to establish a definite correspondence between these alternating fragments of information.



   Any other presentation configuration of tapes of coded records, with regard to the number of lines of said tape, the correspondences of numerical codes and the additional marks such as those defined in the example illustrated by line d of the tape, will obviously come within the scope of the invention.



   Fig. 4 gives a schematic view of the whole of the typing typing apparatus then making it possible to establish a ribbon such as that which has just been specified in relation to FIGS. 1 to 3. In this device, derived from a typewriter - the derivation mode would be similar for a device incorporating part of a teleprinter - for example, and without limitation, two keyboards 3 and 4 are established. pectively associated with two baskets of characters 5 and 6 but with a single roll 7 enlarged to be able to be struck both by the characters of one basket and by the characters of the other. The mechanisms for controlling the character bars using the keyboard keys can be of any standard type and are not shown.

   It will however be noted that for the particular purposes of the invention, any connection is eliminated between the keys of the keyboard 4 and the advance of the roller of the apparatus. Therefore, any service indication grated on a preferably transparent sheet of paper 8, carried on the

  <Desc / Clms Page number 19>

 part of the roll cooperating with the basket 6, will be exactly in the same position of the OM sign character typed immediately after, on the sheet 9, carried on the other half of the roll 7. For simultaneous reading thereafter, it will suffice to apply the two sheets one on the other, by shifting their spaces, which will allow a convenient checking of these recordings in clear.



   It would obviously be possible to provide only one character basket, a single keyboard and consequently a single unencrypted recording. However, it would then be necessary to provide an offset in height of any typing of service information vis-à-vis any typing of text information, so that the service information appears in an interline of the text. This shift could affect the roller 7, by actuating a pawl by a specialized "service" key, any normal typing bringing the roller back to its normal line position for typing text.



   If the service information cannot each be given by a single character, the roll 7 will be divided into two independent rolls in their lateral feeds and the characters in the basket 6 will advance their own roll.



   Specialized keys are, in the example considered, incorporated into keyboards 3 and 4. Their definitions are sufficient for their practical realization,, the reach of any typewriter technician in keyboard 3, the keys T, B, A and F; in keypad 4, keys J and V.



  These keys do not activate characters in the baskets.



  The A key causes the roller 7 to shift by a predetermined number of steps to the left. The F key causes the

  <Desc / Clms Page number 20>

 the automatic coming of the roller to its extreme position on the left. The carriage return can be done by hand, as usual. Blanks will be obtained manually by the space bar E, as usual. An additional end of line hyphen key, U, will be added in keyboard 3 for example. In the body of the text, the hyphen has its place in the normal keyboard 3.



   In the example shown, there is also provided a special key N, between the two keyboards and this key N are associated with ten number keys, from 0 to 9. This subset of keys is intended for entering the numbers of the keys. line of typed text, at least for the preparation of correction records. It can be put out of service for any keystroke if the above-mentioned means are incorporated in the apparatus for automatically printing the line numbers one after the other on the carriage returns. Not being imperative, these means are not illustrated.

   The keys of said sub-assembly can advantageously ensure the typing on the two sheets 8 and 9 at the same time, it is only a matter of a mechanical connection of these keys with the characters of the baskets 5 and 6.



   Any key pressed in one or the other of the two keyboards must ensure on the one hand the advancement of the recording tape, on the other hand its perforation corresponding to the information carried by said key. With the arrangement proposed in fig. 4, the wiring of each key is uniform in principle. As indicated succinctly, the keyboard 4 is arranged so that the pressing of a key corresponds to an indication which must provide the information to perforate therein in a complete column of the tape. In fact then,

  <Desc / Clms Page number 21>

 keypad 4 will have one hundred and twenty-eight keys, numbered from 0 to 127 and the numerical value of the justification will be supplied to the operator in the form of two successive numbers to be typed.



   For the N sub-set, the binary codes to be printed can be determined by an appropriate wiring of the ten keys from 0 to 9: the attendant can then press the N key, when locked (as for a capital letter key), type the line number then press the N key again to release it, and this release will cause the recording tape to advance by two consecutive steps, in which the printing of the two portions of the binary code translating the decimal code typed will take place.



     After pressing the B, F and A keys, the attendant will normally have to press the space bar E. These maneuvers can be omitted by wiring the A and F keys and the space bar E so that their operations remain. pectives cause two consecutive feeds of the recording tape (delayed relay in cascade) and two predetermined perforations as said in these two feeds of one step each.



  The B key could then be omitted. From another point of view, the key T. could also be omitted by wiring each of the keys of the keyboard 3 so as to cause a weak advance of the recording tape and the automatic printing of the announcement T in the first position. not from this advance.



   These keys were however shown and defined because they would be essential in the case of a single keyboard machine. In this machine then, not only would the J and V keys (as well as N) cause the recording tape to advance for printing, but they would also control

  <Desc / Clms Page number 22>

 (similar to what has been described for the N key in the diagram) the introduction of encryption wiring before printing the digital codes associated with them.



   Returning to the example being described, FIG. 5 shows an example of an arrangement for controlling the recording tape and its perforation. In this figure, any key 10 of keyboards 3 and 4 has its support rod 11 pivoted by levers 12 and 13 on pivots 14 and 15.



  When it is subsequently pushed in, two contacts 16 and 17. There are multiple contacts 17 on the control winding of the step-by-step advance relay 18 of the ratchet 19 for controlling the shaft 20 of the ( or) debtor (s) 21, entrained the tape to be perforated 1 on a plate 22 opposite, a row of eight punches referenced j, n, v, t, b, a, f, d to recall their correspondence to the perforations of the tape lines, fig. 1 to 3. Each contact 17; by its closure, therefore causes the tape to advance by one step. The contact 16 of each key is suitably wired to the punch control electromagnets, the windings of which are shown in the diagram in fig. 5.

   The wiring itself does not need to be detailed, given that it will be carried out directly from the desired codes, codes which are of course left to the choice of the user. Closing the contact 16 therefore causes the entry of a numeric column code of the ribbon or of an announcement code, depending on the destination of the key which controls it.



   Although schematic, Figs. 4 and 5 sufficiently inform the producer so that the target apparatus can be directly established, without further indication, given the current state of the art in this field.

  <Desc / Clms Page number 23>

 



   The tape carrying coded records having been produced, it is introduced into a read head included in the input equipment of an electric calculator. This read head can be of a configuration such as shown in FIG. 6; namely to include:
1) A debtor 23 for driving the perforated tape, controlled by the ratchet 24 placed under the actuation control of an electromagnet 25.

   This electromagnet carries an excitation winding 26 and a blocking winding 27 (rendering ineffective any advance control pulse which may be applied at 26 when energized);
2) A read head proper, consisting of eight lamps such as 28, housed in a plate 29 on which is applied the tape 1 driven by the debtor 23, and eight photoelectric cells such as 30, facing the openings established in the platinum 29; any perforation of the strip will give rise to the generation, by photoelectric means, of a signal at the output of the cell corresponding to its row location on the strip;
In normal stepping, each signal output will consist of a voltage pulse;

   the tape will have in certain cases, to be taken further, to park on the plate, each signal then delivered will be converted into a constant potential value during each stopping time;
3) A set of individual preamplifiers, such as 31, each arranged to deliver a high voltage across terminals such as j, n, v, r, b, a, f, d, when a perforation occurs in front of the corresponding cell ; derivation at the output of each amplifier 31 is a polarity inverter stage such as 32, then giving the

  <Desc / Clms Page number 24>

 reverse or complementary voltage, here designated by @, n, v, t, b, a, f, d, according to a notation known in itself (it is read "bar") in Boolean algebra.



   In practice and preferably, the outputs of a read head will be taken from the output resistors of cathodyne stages, in order to facilitate the simultaneous driving of several circuits for using these signals. If necessary, in the various branches, power amplifier stages can then be inserted.



   Any read head existing in a calculator established for the implementation of the invention may have a constitution similar to, or derived from that of Fig. 6, In the example being described, in particular, there will be a second reading head, identical to the first, for the tape of the corrections recordings.



   The general organization of a calculator for any system for practicing the invention can be considered as defined by the block diagram of FIG. 7. It basically presents:
1) An input device E, for reading the coded tapes, text and service and corrections, subsequently incorporating the two read heads 50 and 51 each provided with a step-by-step advance electromagnet 52 and 53 respectively. These electromagnets are both permanently supplied by control pulses coming from an external source, contained for example in the program P of the calculator which fix the application times (most often, this application will be uninterrupted).

   Said pulses are applied to the input terminal 54 of the equipment E.,

  <Desc / Clms Page number 25>

 
A set 55 included in the input equipment overrides the selection of the head to be activated and the operation switching of these heads. An example of the constitution of this set will be described below. By its two output terminals 56 and 57, of voltage conditions always opposite to each other, the selector switch 55 blocks one of the electromagnets 52 and 53 and unlocks the other. However, the two electromagnets can be simultaneously blocked by a program command indicated at 74, which then stops the entry of any data into the arithmetic operator OA of the calculator.

   By means of its two output terminals 58 and 59, always also in reverse or complementary voltage conditions, the selector switch 55 controls the conductivity of one or the other of two conditioned transfer circuits 60 and 61, respectively combined. - These are themselves to read heads 50 and 51 for the transmission of the signals delivered by these heads to a joining circuit 62. By joining circuit is meant here any circuit or network, of the unidirectional type, which transfers any of its input signals to its output.



   In general, the transfer circuit 60 will be made to pass through the selector switch 55 whenever the read head 50 sees its electromagnet advance, except during periods of resumption of reading after corrections, as well as will be clarified with regard to fig. 8. The transfer circuit 61 will be made to pass through the selector switch 55 whenever the signals coming from the read head 51 must be effective.



   The output circuit 63 of the input equipment follows the network or meeting circuit-62. It is in this

  <Desc / Clms Page number 26>

 circuit 63 that can, when necessary, be established the power amplifiers (voltage amplification) of the elementary signals of the codes transmitted to the arithmetic operator.



   2) An arithmetic operator OA comprising a computer circuit 65, connected to the input 64 of the operator by means of a switching circuit or network 66.



  The nature and role of this referral network will be specified later. A certain number of active links from the computer 65 to the program P and vice versa are provided as usual. Details will be given for an illustrative embodiment of said computer.



   By derivations taken from the input circuit 64 of the arithmetic operator OA, the data entered, in part at least, will be written into a memory 68.



  Access to memory 68 is effected through a conditioned routing circuit 67, which will be specified later.



  In the case where the calculator is taken of a so-called universal type, this memory 68 may be all or part of the general memory of the machine.



   Basically, for the desired goal, the computer 65 will have to end the operation in memory, two calculation results. These memories are indicated at 71 and 72, supplied by the outputs 69 and 70 of the computer 65. The data to be entered therein will be defined below. In the case of a universal machine, these two memories may form part of the general memory of the machine.



   Once the two climbs at 71 and 72 have been obtained, the operator will have to work out the output signals. To do this, equipment 73 is provided, supplied by the three

  <Desc / Clms Page number 27>

 memories 68, 71 and 72 as will be said. Its proper operation is placed under the control of the program, link 80. At 75 is indicated the result output of the arithmetic operator OA, coming from the equipment 73.



   3) An output device S comprising at least one writing head 77 controlled by a step-by-step advance electromagnet 78. This head can, if necessary, receive the signals to be written not directly from the input 76 of this equipment S but through a matrix 79 converting the set of signals delivered at 75 and received at 76 into another set of signals to be effectively recorded on a final medium, at least. The triggering of the step-by-step advance of the writing head is controlled from the program P, by the link 83. The entry into service of the converter 79, if provided, is controlled by the program, by link 81.



   4) And, according to the normal routine of calculators of the type in question, program equipment P, the details of which will gradually appear during the specialized description of the machine. However, it obviously incorporates conditioned order units, fixed orders, numerical value repertoires, subroutine commands for the arithmetic operator, etc.



   From this block diagram, it clearly emerges that the calculator to be included in a system making it possible to implement the method according to the invention can actually consist of a machine of the universal type, since a machine of this type necessarily incorporates the various equipment items. and exposed organs. However, since this is an almost fixed program here, a specialized machine can also be established.

  <Desc / Clms Page number 28>

 without going through the organization of a machine of the universal type. It is on such a specialized machine that the presentation will now be carried out and its constitution and operation will confirm that in its place can be used - a universal type. However, to clarify said description, we will consider above all the case of machines working in parallel binary numeration.

   From the point of view of the field of the invention, it should be understood that this in no way excludes any application of the calculation to a machine working in serial binary numbering, the passage from one to the other being obvious to the specialist.



   Previously, however, we will describe in relation to FIG. 8, a possible organization of the selector switch 55 of the input equipment E of FIG. 7.



   In fig. 8, 50 and 51 have indicated the read heads of FIG. 7. The outputs of these heads are, by the conductors shown, connected to a switch, 84 for the head 50 and 85 for the head 51. By these switchers are respectively controlled registers 86 and 87, static for example. With these registers or static recorders are respectively & associated decoders 88 and 89.

   Each decoder supplies at all times a DC voltage proportional to the digital code contained in its register:
The outputs of the two decoders 88 and 89 drive the two inputs of a comparator 96 established so that, as long as the decoding voltages are unequal, and whatever the direction of this inequality, the outputs 56 and 58 present a value potential specific to respectively unlocking the step-by-step 52 of the read head 50 and the transfer circuit 60 associated with this same read head. The

  <Desc / Clms Page number 29>

 The potential values at the outputs 57 and 59 are then such that, on the contrary, the step-by-step 53 and the transfer circuit 61, associated with the other read head 51, are blocked.



  For such an overall condition to be satisfied, it suffices for the potential of the output 90 of the comparator 96 to be directly applied to the outputs 56 and 58, while this same potential has its polarity reversed by the inverter stage of polarity 91 inserted in the link going from 90 to 57 and 59.



   The output of the decoder 88 is on the other hand derived on an input of a second comparator 95. The other input of this comparator receives the voltage coming from 94 but only when this voltage passes through a transfer stage 93 on which it. is also applied. The switchover is only made effective when there is a suitable voltage applied to 93 by the output 92 of the comparator 96. The output of the second comparator 95 is also directed to the output 58 of the selector switch. A meeting circuit or network (definition given above) has been indicated in 97 for the routing of the two voltages at 90 and at the output of 95 to this output 58.



   Operation can be exposed as follows:
Initially, recorder 86 contains a line number of typed text and recorder 87 a number greater than the first. Via the decoders 88 and 89, the comparator 96 delivers at 90 a potential which at 56 and 58 unlocks the step-by-step 52 and the transfer circuit 60 associated with the read head 50. On the other hand, this potential blocks the step-by-step. -pas 53 and the transfer circuit 61 associated with the read head 51. Only the text and ser-

  <Desc / Clms Page number 30>

 main vice under read head 50.



   Each time the text line number changes on this tape, the record at 86 changes. When this record becomes identical to that which exists at 87, the decoding voltages become equal and comparator 96 varies its potential at 90, d ' wherein the unblocking of the stepper 53 and the transfer circuit 61 associated with the second read head and the reciprocal locking of the stepper 52 and the transfer circuit 60 associated with the first read head. From that moment on, the signals transferred to the arithmetic operator are those coming from the correction tape advancing under the read head 51.



   However, in its unequal condition, the comparator 96 blocked the conditioned transfer circuit 93.



  Consequently, the comparator 95 was in a first well-defined condition since it compared to zero (or any other fixed value) the decoding voltage applied to it by the bypass 94 of the decoder 88. When the comparator 96 changes condition, it turns circuit 93 'on and comparator 95 also changes condition. This comparator has. a memory, consisting for example of a bistable rocker with two distinct actuation inputs. When it changes condition, it superimposes at 58 a blocking voltage which, even alone, would be sufficient to block the transfer circuit 60.



  The presence of the meeting network 97 which is unidirectional by definition, however, prevents this voltage from being applied at 56.



   As soon as, in advance of the correction tape, the number of the line at 87 becomes different from that which remained displayed at 86, the comparator 96 operates again.

  <Desc / Clms Page number 31>

 to re-establish the initial conditions: the correction tape is stopped and its transfer circuit 61 is made non-conducting. The main ribbon is restarted. The read head 50 would then deliver excess and unwanted signals to the calculator since these signals represent a canceled portion of text and service. But the comparator 95 has not changed its position and consequently the blocking voltage which it delivers remains applied at 58, the transfer circuit 60 remains blocked. When the main ribbon, restarted, delivers the next line number to register 86; the decoding voltage at 88 therefore at 94 varies.

   It cannot cross the circuit 93 blocked by the comparator 96, but it is applied to the other input of the rocker 95, which restores the rocker to its first condition, hence the elimination of the blocking voltage at 58 The signals from the next main tape line will therefore correctly pass through transfer circuit 60.



   In fig. 9, an embodiment of switching and register has been illustrated, such that it can be provided for the diagram of FIG. 8. In this example, the register comprises fourteen rockers, in two groups from I to VII and from VIII to XIV. Each rocker has two symmetrical and distinct drive terminals, 98 for its reset binary zero condition, 99 for its binary one condition. These conditions are obviously determined once and for all in the assembly of the register.

   A plate output of at least 100 of each rocker is connected to an input 101 of a decoder 102, not detailed because it may have any constitution well known in the art, resistance decoders, diodes, etc. or even relay here since a large

  <Desc / Clms Page number 32>

 speed is not necessary. The only condition to be satisfied is that the voltage at 103 of said decoder varies as many steps as there are units in the number entered in the register.



   This could be considered prohibitive in practice if it was recalled that in fact, the output indicated at 103 may be multiple, as may all inputs of an analog comparator. Any subdivision for decoding can then be conveniently and suitably ensured in the register: said subdivision can be based on a practical consideration, that which statistically indicates an average maximum value (over time) for the difference between line numbers of main text and correction text.



   In this register should be introduced only the line numbers of the typed texts - This is why each entry 99 is preceded by an intersection network, of a usual type with diodes or rectifying crystals on the given diagram, and presenting each two separate entrances!
On an input, is applied the signal coming from the reading of a line of the tape of the machine: the line j for the rocker I, the line n for the rocker II, and so on as shown. A registration authorization signal will be applied to the other input, in the absence of which any signal applied to the first input will remain inoperative.



   This register, on the other hand, must keep the register until a new number is to be entered into it. On the other hand, it must be reset to zero before the introduction of this new number.



   The registration and delivery authorization signal

  <Desc / Clms Page number 33>

 at zero, the same here, is directly derived by the intersection network 104 constituting the entry of the switch (the complete switch comprising this network 104, the networks 105 above-defined and also the separating means which go, be specified and which include the three cathodyne triodes 106, 107, 108 and the delay elements 109 and 110). The intersection network 104 has eight inputs, respectively connected to the outputs j, n, v, I, b, a, f, d of the read head, FIG. 6. Like arrays 105, it is shown to be set for positive pulse signals, each signal being said to be present or being considered to exist when the potential of its application terminal goes high.

   This results from the direction of connection of the diodes, and from the positive polarization common to their anodes, at the point of connection of the sampling paths.



   The first sampling channel, by the separator crystal 111, attacks the gate of the triode 106. The cathode output 112 of this triode is connected to each and every one of the reset terminals 98 of the flip-flops I to XIV of the register. . Thus, each time that the announcement perforation N appears on the read tape, the register will be immediately reset to zero.



   It is easy to verify from Fig. 1 that only the announcement signal N can produce this result, given the combination of signals considered at the input of 104.



   The second sampling path is delayed from 0 to 109, then it attacks the triode 107 by the separator crystal 113. The cathode output of this triode is connected to each and every one of the unblocking terminals of the networks 105 of the rockers I to. VII.If therefore, in the first column of

  <Desc / Clms Page number 34>

 tape following the announcement column N, there is a numerical code, the elements of this code will be in the form of pulses applied selectively to the terminals of the same references as the said perforations of the networks 105, released for the rockers I to VII. This code will be entered selectively bringing into binary condition any concerned rocker.



   The third sampling path is further delayed by 0 by the element 110 then it attacks the triode 108 through the separator crystal 115. The cathode output 116 of said triode is connected to any unlocking terminal of the networks 105 of the rockers VIII. to XIV. This unlocking takes place at the time when the second of the columns following the announcement column N sees its digital code read by the read head. This complementary digital code to the previous one for the constitution of the complete display of the register, will therefore be selectively registered on the group of rockers from VIII to XIV.



   The technology of fig. 9 is well known per se in electric calculating machines. It has been recalled here above all so that it will subsequently make it possible to explain the constitutions of other constituent elements of the calculator, in particular in the arithmetic operator.



   The input equipment E of the machine shown schematically in fig. 7 does not require further explanation for its practical realization. By now considering the arithmetic operator OAn we see that two operating modes are a priori possible, corresponding to two different organizations of this operator:

   the first would consist in introducing into memory in the arithmetic operator a

  <Desc / Clms Page number 35>

 important amount of information, the second, which seems preferable here so as not to burden the production, especially from an economic point of view, consists of entering the data into memory as and when required.
The numerical value of the justification J must of course be introduced first into the machine, and this will result from the ribbon configuration in which this value appears at the top of the records. In a machine according to the invention, it can be modified during the actual text, automatically by the appearance of a new value of J on the ribbon.

   The value J clearly defines that it should be the length of any line of text that may follow it as long as it is not modified.



  However, it is now a line of the text to be composed and not a line of the typed manuscript, therefore a line of coded recording.



   If there were no conditions or restrictions on the possibilities of breaking lines in the text to be composed, the operating program of the calculator would simply be as follows:
A number of Manuscript signs and blanks would be entered as initial data, along with their weight values, T, B and V codes, respectively. By weighting value is meant the number of typographical units to be attributed to any sign and any blank.

   This value may depend on:
On the one hand, from the nature of the variety of character to which it is called;
On the other hand, the value specific to this variety which must be attributed to 1 base unit in which the

  <Desc / Clms Page number 36>

 widths of signs and blanks are expressed;
Finally, the code of correspondence between the nature of the signs and their number of units in the variety considered; all information conveyed here by the digital codes V as defined above.



   It should be noted here that in general the value of a blank is taken constant in usual typography, for a variety of characters, but that the invention allows greater flexibility on this point because such a blank value may , moreover, be varied at will of the operator. This is of course the value of the justifying blank corresponding to a blank of the typed typing after it has been weighted in the machine.



   Having then a set of weighted signs and blanks, the calculator would take from this set a sequence of sufficient length to be at least equal by the sum of its unit values to the numerical quantity 'of justification J, expressed by definition in this same typographical unit. As equality is never obtained in practice, the calculator would take the said sequence by subtracting from it signs and blanks until obtaining the closest value by default of the justification J. Let L be this value.



   It would check how many weighted blanks (or justifying spaces) exist in the sequence retained. Let now hold N this number.



   It would then operate the division (J-L) / N, which would give two numerical values one for the quotient of the division, or Q, the other for the remainder of the division, or R.

  <Desc / Clms Page number 37>

 



   Taking the sequence L again, and by not considering any mixture of engraving fonts having different base spaces, the calculator would then proceed to justify this sequence so that it represents a complete justified line, by adding for this purpose (Q + 1) typographic units with, R spaces of this series and by adding Q typographic units to (N - R) spaces of this series, that is to say in total: R (Q + 1) + (N - R) Q = QN + R J. - L the last equality of this expression obviously defining the operated division.



   However, the calculator must be empowered to justify a line thus cut in a set of signs and blanks only if the cut is lawful. There are, in fact, certain prohibitions on cutting into typographical text.



  These prohibitions can be of two kinds: a) They are linked to signs which follow or precede a cut, the legitimacy of which must be examined; b) They are related to the meaning of the words or propositions which follow or precede a ± break.



   The prohibitions of the first nature are the only ones to be taken into consideration here. Second nature prohibitions, or "meaning prohibitions", could only be dealt with, if necessary, by an intelligent intervention by the preparer and then appear on the tape in the form of a particular cede to be read by the machine and to be interpreted as prohibition.



   As regards the sign prohibitions, namely the prohibitions of the first nature, an exhaustive list can be drawn up in advance and this list introduced

  <Desc / Clms Page number 38>

 in the machine's program memory. To better fix ideas on this particular point, we can consider the following case, limited to prohibitions relating to four letters or signs only.

   By designating the vowels' by the letter V and by the letter C the consonants other than the letters T, X and Y, in addition designating by 5 any Arabic numeral and by S any sign other than a blank , a list of sign prohibitions, for the French language, would be of the following nature:
SV / V s S 5/5 SS = / SS (SX / VSE 5/5 SSS / = S (SY / VSS 5/5 E s (/ sss V / XVE 5/5 E s S /) S (S ' / SS 5 './ 5 5 s S /, S (SS /' SS 5 /. 5 SS /; S - T / - SE 5 /. 5 SS /.

   S
S S / S. 5, / 5 5 S if! S S 5 /, 5 S if? S
E 5 /, 5 S S /: S
I S / S S (S I / S S) (S S / E S
In the above table: S denotes any sign other than a justifying space; V denotes a vowel; C denotes a consonant other than T, X or Y; 5 any Arabic numeral including zero; E designates a justifying space.



   The punctuation marks and the hyphen are represented by themselves.

  <Desc / Clms Page number 39>

 



   In the operation exposed pins above, then, we must provide for the machine an intermediate step which will consist, once de @@ ni3 the aforementioned sequence giving the value closest to J by default, to be examined at the end of this follow-up whether or not there is a cut-off prohibition. If this prohibition existed, the sequence would have to be shortened by the numerical value specific to the signs to be deleted there until the cutoff becomes ilite. Let Lc be the value of said sequence thus rectified, it is with this value Lc that the following operations, as they have been defined, should be executed.



   In the event that it is also wished to take into account the prohibitions-meaning in the calculator, in addition to the information having to appear on the tape of ce- dies as indicated, it would then be advisable to enter this information, when the reading of this tape, in the program memory of the machine and, during a cut-off license operation to involve the content of this memory in the licitation tests, any sum comes into play the prohibitions -signs which are obviously stored in advance then, in the program memory of the machine.



   The general operating program which has just been defined is of course directly applicable only to a continuous text, not having any paragraphs. For the presence of a paragraph, it must be modified as follows, for example:
In the presence of an end of paragraph signal, the sampling of the sequence which has been called L in the set of weighted signs and blanks must be interrupted. The present value of L, i.e. Lf, must be brought to L only by

  <Desc / Clms Page number 40>

 the means of justifying blanks added until the pfecised condition is obtained, namely Lf + Bf (by designating by Bf the value of this series of complementary blanks) equal to the closest value by default of J.



   The appearance of an end of paragraph signal must therefore also control the aforementioned division operation so that this division is carried out according to the relation:
J - (Bf + Lf) = QN + R
The justification must then take place by adding (Q + 1) typographic units to R spaces of the series Lf and Q typographic units to the (N - R) following spaces of this series, then adding the appropriate number of additional blanks. to add the total value Bf to the justified and cut line. It is quite obvious, moreover, that the cut does not need licitation in such a case.



   The presence of a signal at the start of a paragraph, which must ensure the introduction at the start of the cut and justified line of a predetermined number of fixed blanks, i.e. Ba the typographic value of this first series of blanks at the beginning of the paragraph, the appearance of this signal for the start of a paragraph must trigger the execution of the division according to the relation:
J - (Ba + La) = QN + R
The designating the value of the sequence taken from the set of signs and blanks available so that J - (Ba + la) has the smallest possible positive value.



   After the break licitation, the justification will have to be done by first introducing the number of complementary blanks suitable to introduce the total value Ba in this justified line, then normally by adding (Q + 1) units to R spaces of the sequence La and Q units ty @@ -

  <Desc / Clms Page number 41>

 graphs in the following (N - R) spaces.



   By now examining a convenient way in which the systematic test operation for cut-off licitation can be carried out in an automatic calculator, it is seen that it may be advantageous to provide, in the following L and at the end of this sequence , a certain reserve is kL, corresponding to an integer value of a certain number k of signs and blanks. The licitation can obviously be direct, and then the value Lc be in fact equal to (1 - k) L. Now, in the sequence (1 - k) L, as moreover in the total sequence L, the number of blanks is obviously variable and it can therefore sometimes be too low to lead to a satisfactory result. The final typographical presentation of the justified line might be defective in terms of the width of its blanks.



   To alleviate this defect, to a certain extent at least, provision is made according to an auxiliary characteristic of the invention to have recourse to an intermediate operation, before a systematic attempt to make a cutoff. The aim of this operation will be to counterbalance the fixity of the factor k in a sense such that it always minimizes the importance of spaces vis-à-vis the importance of signs in a line.



   To this end, not only will the calculator form the above-defined number N, with an average weighting value of the blanks of the text read, but also a number M which will result from the weighting of these same blanks but with a smaller arbitrary value, which will be considered as the minimum weighting value.



   Before systematic cut-off licitation test

  <Desc / Clms Page number 42>

 the cal @ latrice will make the difference:
M - [(J - L) + kL].



  If this difference has a negative value, the reserve is progressively reduced, sign to sign and white to blank, so that said difference passes through zero. The reserve will then have a minimum value so that after licitation of, cut, the value L obtained ensures, for the justification, a correct typographical presentation in all cases from the point of view of artistic presentation.



   Having thus defined general program conditions for a calculator to be included in a system in accordance with the invention, the description of the exemplary embodiment undertaken in relation to FIGS. 1 to 9 will now be continued by considering fig. 10 which illustrates illustratively a mode of establishment of the arithmetic operator OA of FIG. 7.



  For the convenience of the drawing alone, this fig. 10 is subdivided into two parts, 10A and 10B, with part 10B to be placed to the left of part 10A.



   At the bottom of fig. 10 is recalled program P.



  The links going from the arithmetic operator to the program and from the program to the operator (at least for those which are shown) are shown in phantom. The links for resetting and reading certain elements of the diagram by program signals are not shown in order to simplify the representation. They will be specified, as to their existence and their effectiveness time, during the presentation itself.



   The signals coming from the input equipment E of the calculator are present in the input 64 of the arithmetic operator, fig, 10A. They are respectively applied to the sixteen bounds d, a, f, f, a, a, b, b, v, v, t,, j, J, n, n.

  <Desc / Clms Page number 43>

 



   These signals are directed to a set of routing networks 66. They are also, in part of them, derived to a circuit forming the sub-assembly 67 of FIG. 7. Signals j, n, v, t, b, a; f, d, first of all, are delayed and their derivations by the delay elements 117. This delay is expected to be equal to the duration of the normal time interval of the calculator. Each of these branches results in an intersection network (or conditional transfer) forming part of a set of eight networks, indicated by the reference 118. The transfer control is simultaneous for these eight networks. Their outputs lead to the eight, parallel inputs of a memory, set reference 68.



  This memory can however conveniently be subdivided into two parts. In one 119 will be registered the only numerical value of justification. The content of this memory will not be erased until the introduction of a new datum of numerical justification value. The other part 120 constitutes a shift register, each stage of this register obviously comprising eight rockers in parallel and the shift taking place row by row at each entry of column data. For storage, the offset can be produced by applying a pulse applied at 121 and from the same circuit, which will be described, which controls the conductivity of subassembly 118.

   Part 119 of general memory 68 can then have its independent control, as shown illustratively in the diagram FIG. 10B. The terminals from j to d indicated on the left are considered connected to the terminals j to d, a switching network, which may moreover be similar to that of FIG. 9 is established in 123 for, to allow only the transfer of the

  <Desc / Clms Page number 44>

 fication to the memory 119 which can also be of the type shown in FIG. 9. Of course, in this arrangement, the admission signal j, n, v, t, b, a, f, d) will be formed instead of j, n, v, t, b, a, f, d (in the case of fig. 9).



  Each time this admission signal is formed at 123, and at its three useful times, it will come, via connection 124, to block a subset 125 of three intersection networks.



  This sub-assembly, as we shall see, has the task of forming the admission signals of the codes in the part 120 of the memory and also of forming counting signals to be used subsequently for reading the memory 120, as will be described.



   From entrance 64, fig. 10A, are in fact also derived the signals d, without delay, d delayed by #, and d not delayed, and also d delayed by #. These four signals are combined two by two to form the admission signals of the digital codes in the memory 120. To this end, three intersection networks, subset 125, are provided. The first of these three networks receives the signals d and of. It will therefore deliver a signal whenever the two incoming signals exist, in other words both have their potential high (or low) values simultaneously. The second network receives the signals of and d. It will therefore deliver a signal whenever these two signals exist simultaneously.

   The third network receives the signals d and de. It will output a signal whenever only its two input signals exist.



   These three signals are grouped together by a meeting network 126. This network 126 will therefore deliver a signal each time one of the three output signals of the inter-

  <Desc / Clms Page number 45>

 ', section 125 will appear at its input, indifferently., Whenever network 126 delivers an output signal (or! transfers one of the three above-defined signals), it turns on the eight networks included in the subset 118, therefore ensures the admission of the code which is presented at this time on the children j, n, v, t, b, a, f, d.



   A first, obvious goal of such a storage arrangement resides in the need to enter into memory 120 only the advertisement and numeric value codes of the elements of the ribbon incorporating the advertisement codes V, T , B, A, F, with the exclusion of the elements incorporating the code J, as was explained above, in the case considered, and in the quite general case, the elements incorporating the code of announcement N (which has played its useful role in the input equipment E of the calculator and which must be eliminated in the arithmetic operator).



   Referring to fig. 1 to 3, we see that the arrangement described ensures this elimination. Indeed, for example, the announcement code N, delayed by # in 117, will arrive on the input of the subset 118 at a time when this subset will not be turned on, since none of the three signals of output of the subset 125 will not exist in the above-defined sense: the signal d.de will be zero since de, will be zero (logic product rule); the d signal also; its, two factors being zero; the signal d.de also, its factor d being zero.

   It is easy to verify that, on the other hand, any other announcement code and any other numeric code than those which apply the N announcement code, will be admitted into the memory 120, except for J as said, since their presentation times 4 the entry of 118, one of the three

  <Desc / Clms Page number 46>

 admission signals will always exist.



  If we only consider the first goal of the arrangement described, it would obviously suffice to form only by a single intersection network, an admission prohibition signal (logical product d.de) to achieve this. goal.



   But the second purpose of this arrangement is to allow specialized counting, column by column, of the information stored in 120.



   Any signal - pulse - formed by the logic product d.de will be routed directly to input 127 of a pulse counter 128. Referring to fig. 1 to 3, it can be seen that this signal will correspond, in all cases, to the introduction into the memory 120 of an announcement column information item. Counter 128 will therefore advance one step (or count one unit) for each of these columns.



   The signals formed by the logic products d.de and de.d correspond to the entries in the memory 120 of information contained in digital columns of the ribbon which follow the announcement codes V, T, B, A and F. These signals are directed together on the input of a switch 129. The role of this switch is to direct any impulse that it receives on the input 127 of said counter if this impulse corresponds to the numerical code of a column preceded by by the announcement code V. On the contrary, it will direct any impulse it receives on the input of an auxiliary reserve meter, 130, if this impulse corresponds to the numerical code of a column preceded by one of the aforementioned announcement codes other than V.

   To this end, the dispatcher 129 is placed under the control of a circuit 132 which, each time the announcement code V appears and after a suitable delay

  <Desc / Clms Page number 47>

 (here 2 # since a delay # is provided for the formation of the admission signals and it is the column which follows V), switched the output of 129, of the input 131 of the auxiliary counter 130 at the input 127 of the counter 128. This amounts to saying that the unit 132 forms the logic signal j. not. v, t. b, a, f, d. and that a pulse thus formed is delayed and then directed to the dispatcher 129.



   The auxiliary counter 130 has its output connected to the input 127 of the counter 128. It thus constitutes a reserve of counting pulses, the reserve pulses being transferred from the auxiliary counter to the main counter as soon as the first is filled. In fact, the auxiliary counter 130 could advantageously consist of a shift register, automatic during data entry, controlled as will be described at a later stage of the operating cycle of the calculator.



   To understand the purpose of this arrangement, it is necessary to recall what has been said concerning the licitation of cuts and also the final presentation of the typographic line: after having considered a certain series of signs and blanks, we examine, in a reserve, first if this reservation is not too strong, then if in its content, a cut can be authorized.



   The counting mode proposed above for storing information in 120 corresponds to these two operating stages of the arithmetic operator, acting only on signs and blanks (and independently of the announcement codes and numerical variety codes); the final content of the main counter 128 should depend on the result of said operations to subsequently command a reading and erasure.

  <Desc / Clms Page number 48>

 Partials correct in memory 120 during the actual justification operation. It is therefore advisable to provide for a counting reserve of the same value and of the same nature as the reserve provided for the above-mentioned operations and relating to numerical codes only of signs and blanks.



   It would be possible to make the routing also depend on the announcement codes A and F, in 129, but this will be made unnecessary by the fact that, in the case of an end of paragraph, the two reserves will be completely exhausted before justification and; the case of a beginning of paragraph, this code will come. at the head of a data entry and therefore its count pulse entered at 130 will always be transferred to
127 before stopping data entry.



   Returning now to the input of data into the calculator proper, fig. 10A, we see that the routing assembly 66 supplies six outputs referenced
J, F, A, B, V and T, to recall the natures of the numerical codes that they must deliver. We can consider that these outputs are seven wires for J, V and T, single wire for
B, F, A. In fact, the announcement codes being deleted on our exits, by the set of dispatchers contained in 66 (six dispatchers each of a possible nature and constitution similar to that of the dispatcher who has been described for fig.



   9), are selectively reached registers which, for the writing of B, F and A, will have to have only one element, rocker for example, since no real digital value exists for these signals whose codes known as up to 'now numbers are only marked by a perforation in row d of the column of the ribbon following the announcement column - but who, for the inscription of J, V and? must respect-

  <Desc / Clms Page number 49>

 vement have fourteen, seven and seven elements, such as rockers. The constitution of the six routing networks included at 66 can be deduced directly from that of the routing network of FIG. 9; the constitution of each of the following six registers: 133 for J, 134 for F, 135 for A, 136 for B ,, 137 for V and 138 for T is evident from the above.



  However, it should be noted that while the registers 133 for J and 137 for V will have their resets controlled in the manner indicated for the register of FIG. 9, namely reset to zero only for new display, registers 134, 135, 136 and 138 will be provided to be reset at each code # moment by a pulse from the program.



  When one of these registers receives a piece of information, it therefore only keeps it for a time 9, sufficient for the correct action of the information.



   The codes B and T are also, in the order of their appearance at the output terminals of the same name of the set 66, directed by a reunion network (separation) 139 on the input of a shift register comprising a first register 140 and a second register 141 in cascade with the first. The total capacity of these registers is k + 2, that is to say that they have in all k + 2 recording stages, each stage comprising eight rockers in parallel (seven for the seven elements of code for any letter or sign T, one for the code element for any blank B). Each input line advances both registers by one step.

   The register 14 @. Has a reduced number of stages, four for example if one considers in the machine only the need for four signs to determine a prohibition code.



   By a derivation of the output J, and through a

  <Desc / Clms Page number 50>

 subtractor circuit whose role will be explained below, the justification code is also recorded in a register 143. This register, with parallel inputs for example, is arranged as an algebraic totalizer, that is to say that any numerical code applied on its marked entry (+) causes a record which, if necessary, is added to the entry already contained in the register and that any numerical code applied to its marked entry (-) is' in cut.



  For this, the fourteen rockers of register 143 are connected in cascade in two channels, one for counting, the other for counting down, for example, and a fifteenth rocker, marking the sign of the quantity contained in the register, is incorporated into the structure. In the entry (-) can be included a complement extractor as usual. Depending on the activated input, the up-counting or down-counting channel is released for totalization.



   The sign toggle switch is shown here at 144 output from register 143. From one of these plate outputs there is a connection 145 to the program P. This output can of course present two voltage conditions, one high, the other bass, sign rocker in condition of plus sign and minus sign marking, for example.



   At the start of any cycle for entering data into the calculator, the general condition of register 143 is such that said switch with sign 144 marks the plus sign (including for zero, of the register). At the start of the first data entry cycle, the register is empty and c @ is the numerical value of the justification given at the start of the recording tape which is introduced therein, at the same time as it is in register 133.

  <Desc / Clms Page number 51>

 



   The register 133 contains means suitable for reading it when it is erased (therefore at a change in digital justification value). Then the old digital code, read in this way, is cut off at 142 from the new digital code directed to register 143.



  This register will therefore receive the code for the algebraic difference between the old and the new justification value, which avoids resetting this register to zero during a change of justification during the operation cycle.



   The register 133 contains two groups of seven rockers each, the reading means during the erasure can consist of two groups of seven connections going to the inputs of the subtractor circuit 142, the connections of the second group (the seven rockers of greater weight ) incorporating delay elements @. On a general reset, those flip-flop switches which were in binary one thus deliver code element pulses to their own output connections.



   With register 133 is also associated a reading device 146. This device can comprise here fourteen pentode tubes, the conductivity of which is individually controlled for each by the condition of one of the fourteen stages of register 133 (by the suppression gates for example) . At operating cycle periods which will be defined later, these tubes will receive, from the program, test pulses at 147. On an output link, which may be seven-wire, by subdividing the tubes at 146 into two groups corresponding to the two groups of 133 rockers and alternate test activation from one to the other of these groups, it will therefore appear, in two steps spaced by @, the

  <Desc / Clms Page number 52>

 numerical code entered in register 133.

   The output link is designated by the reference 148 and it leads to an input (multiple here) of a subtractor circuit 149.



   The other input of the subtractor circuit 149 will be able to receive, at times which will be specified and when necessary, a digital code coming from a reading device 150, each time that this reading device is tested by program pulses at 151. The device 150 is intended to read the digital code which can be entered in the register 152. The program pulse reset entry of this register 152. has been indicated at 153. will be able to display will be transmitted to it from the program P via link 154. Their value can be either Ba or Bf according to the definitions of these quantities which have been given above. This register will therefore only come into action in the case of an end or beginning of a paragraph.

   Its constitution, except for the number of rockers, may be similar to that of register 133 and that of reading device 150 to that of device 146.



   The command to display a Ba value or a Bf value in the register 153 by the link 154 is obviously carried out when the condition changes the rocker 135 recording a start of paragraph signal or the rocker 134 recording a end of paragraph signal. This is why the output of the rocker 135 is shown connected to the program P by a link 155 and the output of the rocker 134, by a link 156. It has been indicated at 157 and 158, respectively, the reset terminals of these two. rockers. These terminals receive an uninterrupted series of cadence program pulses #, slightly ahead of the output instants.

  <Desc / Clms Page number 53>

 of pulses of the routing sub-assembly 66.



   It may be recalled here that this switching sub-assembly 66 introduces a delay # at any output, so that the entry of data into the calculator coincides with the entry of data into the memory of the arithmetic operator.



   Any Ba or B digital code appearing on link 154 will also be directed by reunion circuit 159. on the input (-) of the algebraic totalizer 143.



   Any code developed by reading register 152, at 150, will be directed, in addition to the aforementioned input of subtractor circuit 149, to an input of an adder 160. The output of this adder 160 is directed, through a control circuit. conditional transfer 161, on the input (+) of the algebraic totalizer 143. This transfer circuit 161 will only be turned on when it receives in 162, from the program, unblocking pulses, at times which will be specified more far.



   The three registers 136 (a rocker periodically reset to zero by the cadence control pulses # with a slight advance over the pulses which can exit in
B of the switch 66 and applied in 163), 137 (seven rockers returning to rest on each new entry of data to be substituted for the previous one), and 138 (seven rockers returned to rest by applying 164 pulses periodicals of rate 9 and having the phase advance indicated above) have their outputs connected to inputs of two networks or matrices 165 and 166. Matrix 165 receives the signals coming from 136 and 137. Matrix 166 receives the signals. signals from 137 and 138.

  <Desc / Clms Page number 54>

 



   Each of these matrices ensures at least one weighting, in the sense which has been defined above, of the digital codes T on the one hand and of the blank signals B on the other hand.



   This weighting is of course placed under the control and control of the digital V codes. In fact, the matrix
165 is double since it must deliver, at its output 167 the weighted blanks with the minimum basic value in the desired typography, and at the output 168 the same weighted blanks with the maximum admissible value of the typography.



   Either of these values may depend on the digital code then present in register 137; hence the flexibility of the machine in this regard. If this advantage was not deemed useful, the link between 137 and 165 would be omitted and fixed weighting values introduced in 165.



   Each of the aforementioned matrices therefore has the role of converting an incoming digital code (coming from 138 for the matrix 166, coming from 136 for the matrix 165) into another digital code, a function of the first one, and representing, for example, the result of the multiplication of the first by a numerical value fixed by the quantity recorded in -
137. Static constitutions of such matrices are well known in the art of multiplication and function tables for electronic calculators.



   The output 167 of the matrix 165 which will finally deliver the aforementioned number N at the end of each data entry cycle, attacks a memory 170. This memory is established in the form of an algebraic totalizer, for example of the type specified about the algebraic totalizer
143. Signals from 165 are. applied to the input (+) of this totalizer 170. The rocker switch has been indicated at 171

  <Desc / Clms Page number 55>

 sign of the totalizer and at 172 an input of the output to zero by the program.



   The outputs 168 of the matrix 165 and 169 of the matrix 166 are directed, by the union circuit 173, on the one hand on the input of a shift register 174, on the other hand, on an input of the network. reunion 159. The signals they deliver will be, from the latter introduced, on the input (-) of the algebraic totalizer 143.



   Register 174 is of a type of shift register which has been previously described in the discussion. It is planned on k floors. Its shift control, during data entry is provided by terminal 175, simultaneously with the shift control of register 140-141 then ensured by terminal 176; these two terminals will, of course, receive step-by-step advance pulses, of rate #, coming from program P.



   The output 177 of the register 174 is directed to an input of a code adder circuit 178, the other input of which receives the signals coming from the output 167 of the matrix 167. Thus, during a data entry cycle , and as will be specified later, the totalizer 170 will be actuated to progressively form the sum M + (1 - k) L. It will therefore suffice to introduce therein, in negative, the justification value J so that the quantity contained in this totalizer is as defined above.



   The sign rocker 171 of the totalizer 170 controls the conductivity of a conditioned transfer circuit 179, so that this circuit is on when said rocker marks the minus sign and not on.

  <Desc / Clms Page number 56>

 that the sign rocker is caused to mark the plus sign (it will mark this sign as soon as the totalizer register 170 is brought to zero). Circuit 179 will receive, at times which will be specified, program pulses via link 180.



   Any program pulse, transferred by circuit 179, will be applied to the input of a union circuit 181 whose output drives a forced shift command (without input of data) of both registers 174 and 140-141.



   The other input of the joining circuit 181 is connected by a link 186, polarity discriminator for example, to the output of a rocker plate 182. This rocker by the output of its other plate is also connected to the program P by the link 183. This rocker 182 is intended to mark the result of the sign prohibitions tests applied by the program, in 184, on a comparator 185 which compares the codes of these prohibitions with the codes contained in 141. At each coincidence, the rocker 182 switches twice and therefore delivers a pulse at 186 to the meeting network 181.



   If there is no coincidence, the cut will be authorized as explained later, and the rocker 182 will come in a fixed condition which, while only delivering a pulse of wrong polarity at 186 (no return to the rocker), will then apply a voltage to the program in 183 notifying it of the licitation of the cut. Of course, any other organization of comparator and comparison indicator can be used here. For example, the comparator 185 may consist of an adder or code subtractor and the member 182 then detects the output zero of the adder,

  <Desc / Clms Page number 57>

 whereupon it commands the opening of pulse transfer circuits to 181 if there is not this zero condition, to the program if this zero condition exists.



   In addition to the shift control of registers 174 and 140, any pulse leaving the network or circuit 181 is also directed by link 187 to the shift command input of the same reference in FIG. 10B; for the auxiliary meter 130, of capacity k.



   The output of register 174 is also directed, by link 188, to the input of a counter or totalizer register 189, FIG. 10B. During a reading with erasure of this counter 189, ensured by pulses applied in 190 and coming from the program at a time which will be specified, the output of said counter is transferred, via link 191 to the input of adder 160 of fig. 10A, therefore, by this circuit, at the input (+) of the algebraic totalizer 143.



   The content of account 189 can however be read at will in that a reading device 192 is associated with it, the activation of which is placed under the dependence of program pulses to be applied at 193 to this reading device. 192. The output link 194 of this reading device is brought to an input of a subtractor circuit 193.



   The other input of this subtracter, is, by the link 195 and a conditioned transfer circuit 196, connected to the output of the subtracter 149, fig. 10A. In 197 is indicated the activating binding of 196.



   The output of the subtracter 193, fig. 10B, is connected to the input (+) of an algebraic totalizer 198. The sign toggle of this totalizer is indicated at 199. As long as it is in a plus sign condition, it blocks a circuit on the one hand.

  <Desc / Clms Page number 58>

 conditioned transfer 200 and on the other hand turns on a conditioned transfer circuit 201 which will receive program pulses in 202 according to a modality which will be specified.



  As soon as it reaches the minus sign condition, it will pass circuit 200 and block circuit 201. Furthermore, via link 203, it will indicate to the program an end of operation (in fact of division).



   The transfer circuit 200 receives on its input coded reading trains (in parallel for example) from a reading device 205 associated with the totalizer, each time this device 205 receives a reading pulse at 206. However, this reading device reading will in fact consist of an additional register into which will be transferred by a circuit 207, the content of the totalizer 198 each time. that a transfer admission pulse will be applied at 208. Any pulse at 208 will follow in time a pulse at 206, the latter emptying register 205. Thus each signal transported by link 209 from register 205 to the input of circuit 200 will in fact be delayed by one step of. operation on the content of the totalizer.



   The output 69 of the circuit 200 reaches a memory 71 and, by the preceding arrangement; the remainder R of a division carried out by successive subtractions in 198 will be routed to said memory 71.



   The input (-) of the totalizer 198 will receive, for the. division, by link 210, the digital content of a register 211, FIG. 10A, This register has its input connected to the already specified output 168 of the matrix 165. At the end of a cycle for entering data into the calculator, it will therefore contain the number N corresponding to the global value of the blanks

  <Desc / Clms Page number 59>

        weights with the aforesaid average value, which have exists: in these data. The link 210 will receive the digital codes from a reading device 212 associated with the register 211, each time a reading pulse is applied at 213 to this reading device.



   The output 70 of the conditioned transfer circuit 201 of FIG. 10B is connected to the inputs of two memories.



  They together constitute the memory 72 of FIG. 7, then assigned to the entry of the quotient of the division. The memory 214 will contain at the end of the division the number Q, the value of the quotient. The memory 215 will similarly contain the number Q + 1, being arranged to be reset in the marking position + 1, while the memory 214 is arranged to be reset in the marking position 0. Their reset inputs are indicated at 216 and 217, respectively.



   The memory 71 of the remainder R can be progressively emptied, by its down-counting input 218. A zero indicator 219 is associated with it, which varies its output voltage depending on whether or not it indicates the zero of the contents of the memory 71. By its output voltage, direct at 220, reversed in polarity at 221, it controls the conductivity conditions, which will then always be reciprocal, of two conditioned transfer circuits 222 and '223. The circuit 222 will receive for transfer a signal of Q units from the reading of the memory 214 by any read pulse which may appear at the read input 224 of this memory.

   The circuit 223 will receive for transfer a signal of Q + 1 units from the reading of the memory 215 for any reading pulse applied at 225 to this memory.
In fact, the pulses at 218, 224 and 225 pro-

  <Desc / Clms Page number 60>

 will come from a circuit 226 whose function is to detect the existence of blank signals in; the read signals from the memory 120. These signals come from a circuit or dis. positive reading 227 of this memory. Circuit 226 detects blanks in ayamt then, applied in. permanently on sound, entry 228, the code representing a blank {announcement B + perforation d in the following column). and by receiving the codes read at 227 on its other input.

   Each detection of a blank causes in 226 the generation of a pulse applied in 225,224,218.



   Furthermore, the output link of the reading device 227 also reaches an input of an adder device or circuit 229 which receives at its other input any signal passing through circuit 222 or circuit 223. At each indication of blank (perforation) d), this circuit will therefore add a numerical value Q or Q + 1 (perforations in the other lines of the reproduced tape).



   The output link of the adder 229 is directed to a union network 75 constituting the output of the arithmetic operator, FIG. 7. The other input of this network or meeting circuit receives the justification code each time the memory 119 is read by means of a series of pulses (three for parallel reading) applied in parallel. 230 on the reading device 231 associated with the memory 119.



     The reading of the device 227 associated with the memory 120 is ensured by the means of pulses applied at 232 and ensuring, in addition to this reading, the erasure of the data read in the memory 120 and also the counting of the contents of the counter 128. A return to zero indicator 233, associated with this counter then delivers an end of reading signal

  <Desc / Clms Page number 61>

 in the program, by the link indicated in 234.



   The device 227 can consist of as many groups of eight pentode tubes as there are stages provided in the counter 128 and the memory 120. The stages of the counter control said tubes by their suppression grids, the stages of the memory of the devices. control via their screen grids and the read pulses are applied to the control grids, for example, and without limitation. The pulses at 232 consecutively test the stages, in the direction from left to right in the diagram of fig. 10B.



   Having thus specified an example of the constitution of an arithmetic operator for a calculator to be incorporated into a system in accordance with the invention, its operation can now be explained as follows:
The memories being empty in the calculator, the reading of a tape of coded recordings begins with the introduction of a digital justification code in 133 and 143.



  The code is applied at 143 through 142 which then subtracts zero since register 133 is empty.



   On the tape, after the indication of a line number which we do not have to take into account at present, comes an announcement code V followed by the numerical value of V. We consider for the moment a text without indentation. This digital V code is entered in register 137 and will remain there until a new V code is given on the tape. The condition of register 137 then conditions the two matrices 165 and 166
T codes and B codes can then alternate irregularly. Each of them is introduced into the register which corresponds to it, 138 or 136. The values of these codes are weighted in the matrices 166 and 165 according to the code

  <Desc / Clms Page number 62>

 displayed in register 137. This weighting varies each time this V code varies.



   The weight codes which exit at 169 and 168 are directed by the meeting network 173 and the meeting network 159 to the input (-) of totalizer 143. They are also directed to the input of reserve register 174 in this form. weighted and, simultaneously, by the union circuit 139, on the entry of the register 140-141, in unweighted form.



  * These weighted digital codes are thus gradually subtracted at 143 from the digital justification code J. At any time, in this totalizer 143, the value of the current digital difference (J-L) will exist.



   On the other hand, the weighted codes exiting at 167 are directed through adder 178, on the (+) input of tmtaliser register 170.



   As soon as the reserve 174 is filled, the outgoing weighted digital codes are sent to adder 178, so are gradually added to totalizer 170. They are also sent to the entry of register 189 of FIG.



  10B, and at any time then, this register 189 contains (l-k) L.



  At any time, register 170 contains the digital value M + (1 - k) L.



   As the data continues to be entered, the algebraic totalizer 143 passes through zero. Its sign rocker operates and delivers at 145 to the program P an end of data admission signal. The program delivers at 74 a blocking signal for the step-by-step advances of the read heads of the tapes. Data entry ends for the considered cycle of operations, regardless of the head

  <Desc / Clms Page number 63>

 tape reading then in service.



   Due to the delay # in entering data after reading the tape, a last code can be entered in the computer, half of the time it is put in memory 120.



   The program P then stops the application of the shift control pulses at 175 and 176. On the other hand, it applies at 147 and 151 a suitable number of read pulses to the memory readers. 133 and 152.



   Memory 152 is currently empty. The code delivered by this reading will therefore be zero @ and only the justification code J will be directly transmitted through the subtracter 149. Applied to the input (-) of the totalizer 170, this code therefore brings the content of this totalizer to the value M + (1-k) L - J.



   The program then applies a threshold of k pulses at 180. If the result of the totalization which has just been indicated is negative, the sign-rocker 171 of the totalizer 170 turns on the transfer circuit 179.



  The first pulse at 180 will pass through 1 (transfer stage 179 and cause registers 140--141 and 174 to be advanced one step). The latter will deliver a digital code of one sign at a time to register 189 and to totalizer 170 If this digital addition does not vary the sign by the above quantity in this totalizer, the sign rocker will remain in a minus sign condition and circuit 179 will remain on, A second pulse in 180 will repeat the operation, and so on. then, until the content at 170 passes through zero, whereupon circuit 179 is blocked.

  <Desc / Clms Page number 64>

 



   Of course, in normal operation, only part of the k program pulses at 180 will pass through circuit 179, otherwise the number k would have been chosen incorrectly.



   After these k pulses in 180, the program ensures the comparison in 185 between the groups of four codes present at 141 and the groups constituting the sign prohibitions.



   For a coincidence of configurations between these groups of codes, preventing cut-off, the rocker 182, as aforesaid, sends a pulse at 186 which, being of suitable polarity, controls the advance of registers 140-141 and 174 by one step. The latter delivers a one-sign digital code signal to register 189. It also delivers the autotaller 170, but this does not matter.



   As soon as a non-coincidence is detected, the cut becomes lawful. This licitation causes the program to send to the program at 183 a signal for the end of the systemic cut-off test operation. By the su @ - step-by-step advance control described for registers 140-141 and 174, the first at (k + 2) stages in total and the second at k stages, the licita- tion of a cut will find the cohtenu of the last two stages of 141 transferred in its weighted form in 189. The latter then effectively contains the numerical value of legal cutoff L.



   The signal for the end of the licitation operation, at 183, then triggers the mechanism of the division which, in the example considered, will be carried out by successive subtractions of the content of register 211, namely N, from the value digital J - L.



   However, before describing this

  <Desc / Clms Page number 65>

 division, that any offset pulse which passed through the union network 181, either as a result of the k program pulses at 180, or as a result of the pulses delivered at 186 during the search for a lawful break, advanced the counter 130, fig. 10B, so that the content corresponding to this number of pulses has been transferred into the counter 128. The agreement between the content of 128 and the number of stages of the memory 120 which must be read at the end of the operation was therefore well insured.



   That said, the division is carried out as follows: the program P firstly performs a simultaneous reading of registers 133 (digital quantity J), 152 (zero here), and 192 (digital quantity Lc). From the subtractor 193 outputs the digital quantity J-Lc which is introduced by the input (+) in the totalizer 198. Thereupon, the program performs a first reading of the register 211, and it will repeat the periodic application of d 'read pulses at 213 until it receives an end of division signal.



   By the input (-) of the totalizer 198, therefore, the number N is subtracted a sufficient number of times so that 1 toggle of sign 199 of this totalizer indicates zero at least. Prior to each reading of N, the program transfers the contents of the totalizer 198 to the reading device 205, after the first reading of N, At each reading of N, it also applies a pulse in 202. After the first transfer of N and before the second, it applies an impulse at 206, and repeats this operation on each transfer.

   Thus the quatient Q is formed progressively, both in memory 214 and in memory 215 (where the number Q + 1 is formed in parallel) 2-1, At the end

  <Desc / Clms Page number 66>

 operation, the remainder R of the division is transferred to the) memory 71 by the circuit 20Q then turned on. The change of state of the rocker 199 having stopped the operation, no other pulse relating to the division is sent by the program.



   On the other hand then, the program applies the appropriate number of read pulses in 230, from where the output in 75 of the numerical code of the justification J, contained in the memory 119. If the justification must not be repeated until If it does not vary, the output of the justification reading will also be directed to a particular memory in the program. At each subsequent reading period, the program will compare this stored code with that of the new reading and, if the same, will block output circuit 75 for the desired time. Alternatively, the program would independently test the contents of 119 and apply the pulses at 230 only if the result of that test shows a numerical value variation of the justification.



   Then the program applies in 232 the read pulses of the memory 120 until it receives via the link 234 the indication that the counter 128 (which has counted down under the action of these pulses) has returned to zero: In the 'interval, the digital codes of the cut line are output on link 227 for blank test at 226. At each blank signal, a pulse is delivered by 226, is applied to memories 214 and 215 and simultaneously to the input of countdown of the counter 71. During the first R blanks, until this counter has been brought back to zero, the transfer circuit 223 is conducting ;. so that is the value

  <Desc / Clms Page number 67>

 numeric (Q + 1) which is added to the codes of these blanks in 229.

   During the other blanks, once the counter R has returned to zero, it is circuit 222 which is on and these (N - R) blanks will therefore only be increased by Q 'units in 229. Circuit 229 can receive a program from the program. base numeric code each time a blank is detected in 226.



   The cut line justification end signal, arriving at 234 in the program, stops the application of the pulses at 232. The program then delivers reset signals from the arithmetic operator in the data entry repeat condition, to know :
1) By application of pulses in 190, the counting of register 189 takes place and the delivery of the digital code of its content on link 191; therefore, the content Lc of this register is applied to the (+) terminal of the totalizer 143 through the adder 160 and the transfer circuit 161 then turned on. The register 152 is reset to zero if it was not already there and a last reading carried out in 150. Under the conditions considered (text without paragraph) the code Lc will pass alone to the totalizer.

   In 143, there will therefore remain the difference between the digital value J and the residual number of weighted codes which have remained in 174;
2) By shift command of 174 (hence 140-141), the residual content in 174 is transferred to 189; simultaneously the counter 130 sees its residual content transferred to 128;
3) The registers and totalizers 211, 170, 198, 205, 214 and 215 are reset to zero,
The program circuit then removes at 74 the

  <Desc / Clms Page number 68>

 blocking voltage of the read head electromagnets and the cycle begins again.



   In the event that an end of paragraph signal appears during a data entry period, now the described operation will be changed as follows:
The signal sent to the program by the link 156 coming from the rocker 134 for writing the end of paragraph code, will by itself stop the entry of data (output at 74 of an end of reading signal); the following code will be read but not transferred to the matrices 165 and 166 (it is the code of the column which follows the end of paragraph signal and therefore from the point of view of the calculator described, a column N, without action in the arithmetic operator).



   The program then triggers the sending of pulses on link 154, more precisely, the repeated sending of blank codes justifying the end of paragraph. These codes are of course entrenched in the totalizer 143 at the same time as they are progressively stored in memory in 152.



   When the sign toggle 144 changes condition, it delivers a signal at 145 to the program P. This signal is then interpreted there as a command to end the sending of justifying blank codes at 154. The number Bf has thus been formed and put in. memory.



   In the following operations then described, this value Bf will be automatically subtracted from the value J when it is transferred to register 170 by the subtracter 149. The sign of the content of this register will thus be automatically negative and the k pulses coming from the program in 180 will pass through circuit 179. The entire contents of reserve 174 will be transferred to register 198. Register

  <Desc / Clms Page number 69>

   140-141 'will be emptied. The reserve in 130 will be transferred to 128 for reading from memory 120.



   The cut will then be directly authorized.



   The justification of the cut line will obviously have to take place on J - Bf. This is ensured by the fact that when the dividend is introduced into the totalizer 198, transfer circuit 196 made passing through the program at this instant, the number Bf will be automatically reset in 149 from the number J.



   The justification will then operate as said: The end of the justified line can be constituted by the end of paragraph signal, to be interpreted as such in the output equipment S of the machine. The justifying blanks can be added from the output at 75, by placing an end of paragraph signal detector in parallel with the blank detector 226 and by controlling by this end of paragraph signal detector, the addition of validating blank signals in number defined by the contents of memory 152.



   A start of paragraph signal will then appear. always with register 198 empty and also counter 128.



  In the totalizer 143, of course, the number Bf a. was added by adder 160 at the time of Lc transfer which has been described.



   The coming to work of the rocker 135, recording a start of paragraph signal, causes the program to send a start of paragraph signal via link 155. Then, in response to this signal, the program momentarily stops l entry of the data but sends a predetermined number this time of fixed blanks (for beginning of paragraph) in the memory 152 and therefore, by the circuit 159, in the totalization.

  <Desc / Clms Page number 70>

 algebraic tor 143 (by its entry (-) of course). then the program removes the blocking signal at 74 and data entry resumes.



   The rest of the operations take place similar to what has just been described for the end of paragraph signal, i.e. the justification will take place on J - Ba, after the totalizer 170, for l operation prior to the cut-off licitation test, aura.received - (J - Ba). However, the cut-off license will be carried out in the normal way.



  Then, for the division for justification, the dividend will be J - Ba.



   When reading memory 120 for justification, the paragraph start signal will appear at the top. Either it will be directly transmitted at 75, to be interpreted as such by the output equipment S, or else a specialized detector will also be provided in parallel with 226 and will order the sending of the appropriate number of justifying blanks in 75, then temporarily stopping the reading of the memory.



   It would of course have been possible to provide a single signal to indicate the start and end of a paragraph, which would imply the automatic replacement after the cycle containing the end of paragraph signal, of Bf by Ba in the memory 152 and the Immediate subtraction of Ba in the totalizer 143 before resuming the reading of the tape. However, since a text generally begins with a start of paragraph signal, a cycle of operations would be lost.



   From the examples which have just been described, the means to be implemented to satisfy the method of the invention sufficiently emerge so that the resulting devices can be established directly by the technician.

  <Desc / Clms Page number 71>

 



  Of course, any variation in the technological implementation of the means described falls within the scope and the field of the present invention, as they are defined by the preceding description.



   It is obvious that the punch 77 can, by a suitable configuration of the die 79, either directly produce any tape of the type already known for the automatic control of a mechanical composition machine, or produce a special tape to be converted, at the same time. will of the printer, in such a direct order tape.


    

Claims (1)

ABSTRACT. A composition system for printing texts, based on a new process of preparation, cutting and justification of the typographic lines to be printed, which process is essentially characterized in that it comprises the following operating stages: a) Establishment at least one coded recording of the information elements contained in a text and of the service information elements accompanying them, a recording readable by automatic reading; (b) Gradual conversion of this coded recording into electrical signals; c) Transposition of these signs into a set of signs representing, typographical line by typographical line, the complete composition of these cut and justified lines; d) Recording for mechanical or visual playback, or both, of this set of transposed signals. Variants and additions to this method and to this system to allow maximum automation, and in particular the use of an electric calculator for stages b to d defined above. New industrial products made up of the equipment established and intended, in all embodiments, for the practice of this process and this system, as well as all coded records relating thereto.