Go to top of Babbage Technical Description

Stereotype Table


Drawing References: BAB [A] 147, 163, 164, 166

The stereotype table is the apparatus for moving the trays containing soft material (papier mache ('flong'), or soft metal) into which results are impressed. The soft material is then used to make the plates from which tabulated results are printed. The printing and stereotyping apparatus is shown on 163 left. The printing apparatus is the box-like mechanism bolted to the left hand vertical frame members roughly halfway up the engine. The stereotype table is the apparatus immediately below with the characteristic wishbone struts stiffening the cast leg supports.

The result of each calculating cycle is set up on the small and large stereotyping sectors each with thirty type wheels. These are shown in detail on 172 and in outline at the top of 147. The two sets of stereotyping sectors are lowered together to impress the result into the material in the trays below to provide two sizes of type. The stereotyping sectors have vertical and rotational degrees of freedom only but have no freedom to move in the horizontal plane. The position of the result on the page is therefore determined not by lateral motion of the heads but the position of the tray below the heads. The stereotyping apparatus is in effect a programmable X-Y platform which repositions the trays under the stereotyping heads each calculating cycle to receive each new result.

The format of the tabulated results is programmable. The apparatus provides options for line-to-line printing in one or more columns, or column-to-column printing. The number of lines per page and line separation are also programmable. The various layout options are selected by choosing pattern wheels. One pattern wheel, selected from a choice of four, determines line-to-line spacing; a second pattern wheel, also selected from a set of four, controls column-to-column movement The pattern wheels are shown on 163 and 164 extreme left, and in greater detail in 166.

The stereotyping table forms part of the control mechanism of the calculating drive. When a page is completed in the format preset by the pattern wheels, a weight is released into the cylindrical cup shown in 163 bottom left. The cup is connected to the scoop lever of the main drive clutch above the cam stack at the far right of the engine via a cable guided by pulleys. The clutch disengages the drive and the handle runs free. Uncoupling of the drive at the end of a page halts the engine without any overrun and ensures that the integrity of the figure wheel settings and control mechanism when the calculation run is resumed. The automatic halting of the engine at the end of a page is to allow the material in the trays to be renewed. This is done by fixing fresh trays, prepared with suitable material, on the carriage.

The stereotype apparatus is wider front to back than the calculating section (164) and the cast stand for the stereotyping table rests on the base plinth i.e. the legs are well clear of the long framing members supporting the vertical framing pieces.

Description of Operation

The material receiving the impressions from the stereotyping sectors is held in trays called matrix pans. Separate matrix pans for large and small type are shown at the top of 147. The matrix pans are fixed to a travelling frame or carriage (147 top right). The flong needs to harden before being removed from the pan. The pans therefore need to be removable to allow replacement pans, loaded with flong, to be fixed so that the calculation run can proceed while the impressed material is drying. No means of fixing the pans to the carriage is shown and no means of spacing the relative positions of large and small pans is indicated. However, the end view (147 top left) shows a flush fit along the left and right page edges and the Elevation shows the same top-of-page clearance for both small and large pans. The position of the pans on the carriage is therefore fixed left-to-right by the width of the frame and the assumption is that the pans are wedged in place top-to-bottom (page-wise, front-to-rear as viewed from the front of the engine) in accordance with contemporary printing practice.

The carriage runs on two pairs of inverted v-shaped runners. One pair gives the freedom to traverse column-to-column on any given line ('v's shown in section in 147 End View); the other pair, at right angles to the first, gives the line-to-line freedom down the page ('v's shown in section in 147 Elevation). The two orthogonal sets of sliders provide the necessary degrees of freedom in the X-Y plane. The line-to-line motion of the pans corresponds to right-to-left motion of the carriage in the End View (147 (top right). Facing the engine this corresponds to a front-to-back motion i.e. the pan retreats as the table progresses down the page. The column-to-column motion corresponds to right-to-left motion of the carriage in the Elevation (147 top left) i.e. the carriage traverses right-to-left as viewed facing the front of the engine.

The carriage supporting the two matrix pans is shown driven in two separate ways. For stereotyping with large type the carriage is driven by a pair of racks and gears shown in the three views in 147. The two gears (5B, 5P) keyed to drive shaft AA, engage with racks 1R1, 1R2 which are fixed to the underside of the carriage. Incremental rotation of the shaft provides the line-to-line advance of the platform. When stereotyping with small type, the carriage is driven by a single pinion 5a, which engages with rack 4R. The reduced pitch of the pinion ensures smaller line separation for the small type. No means of fixing the single central rack to the underside of the carriage is shown in either 147 or 163. This is an incompleteness in the drawing. The gears for the double rack and the pinion for the single rack are keyed to the same drive shaft. Since the carriage for both matrix pans is a single assembly it is clear that the drives are mutually exclusive: if both were engaged there would be a contention between the different pitches of the two incremental advances. It is clear from this that stereotyping in both small and large type simultaneously (i.e. in the same run) was not intended. Only one type can be used at a time and the drive for the unused size would need to be disengaged. Again, there is no indication of how this is to be done. The pan for the smaller type is shorter than the large pan by roughly the same factor as the difference in pitch between the pinion teeth and the pitch of the gear teeth. This indicates that the intention of the smaller type was not to produce more lines per page but to provide an alternative page size for the same run of results. The smaller type does however does provide the option of up to three-column layout compared to a maximum of two columns in the larger type.

The pinion is centred between the side frame members by two sleeves which act as spacers. The pinion and the sleeves have travelling keyways which slide on the long key fitted to the drive shaft. The whole assembly traverses during column-to-column motion (up-down motion in 147 plan). The pinion is shown wider in 163 than in the two views in 147. This is a drafting inconsistency.

The control mechanism for the stereotyping table is shown in the three views in 166. The two sets of pattern wheels (3N, 6N, ) are shown as the large wheels immediately below the two horizontal bars in 166 (top left view). This -right hand pattern wheel determines line-to-line motion (3N); the left hand pattern wheel (6N) determines column-to-column motion. The left and right hand pattern wheels jointly determine end-of-page action via the cluster of levers and catches positioned between them in the elevation in 166. The motive power for driving the pattern wheels during stereotyping is provided by weights A (column-to-column) and B (line-to-line) suspended over pulleys on the pattern wheel shafts (166 Plan and elevation). The intermittent release of the pattern wheels to advance the carriage and the end-of-page action are controlled by the two horizontal bars E and F (166). The drive to rewind the weights is derived from the main drive shaft (which rotates continuously during calculation) via a dog clutch shown as part of the circular assembly at the bottom centre of the end elevation in 166. The rotation of the line pattern wheel during stereotyping is discrete, with the size of the steps determined by the spacing of the teeth. The rewind action is a smooth anti-clockwise rotation.

The drive for the control mechanism derives from a pair of conjugate cams (4n, 4m) which are driven directly from the main drive shaft, on the underside of the engine (163). The contact cam-followers 5E1, 5E2, and rocker shaft assembly converts the rotary motion of the cams into reciprocating motion of the lever 5E3 (166 elevation). The rounded head of the lever is trapped in a recess in the lower oscillating bar (E) and the bar executes one linear reciprocating oscillation each calculating cycle. The rise and fall of the cam profiles occupies 29o of the 360o cycle (13o each way and a 3o dwell) i.e. about 12% of the cycle. With an average cycle time of six seconds, one outward and return excursion (13/16" stroke) of the oscillating bar takes about 0.7 sec. The bar is therefore stationary (to the right) for most of the cycle.

The incremental intermittent motion for line-to-line motion is produced by the escapement action of the oscillator bar and line catch (1B) on the line pattern wheel. A peg in the lower oscillating bar operates the line bar lever during the outward (right to left) excursion of the bar. This releases the line catch which frees the pattern wheel to rotate. The pattern wheel is driven clockwise by the suspended weight, B and rotates by one tooth for each release.

The restoring force for the line catch is provided by a leaf spring fixed to the back plate and bearing on a lug on the line catch shaft (shown as dotted square just below notation '6B' in 166 plan). The motion of the pattern wheel steps one position clockwise each cycle. This is transmitted to the carriage drive shaft 7A. (147) via a gear train. One gear (3B) is mounted on the pattern wheel shaft (plan view 166). The second gear is shown in outline only drawn concentric with the line catch pivot in the end view of 166.

The line-to-line height is determined by the pitch of the pattern wheel teeth for which no detail is provided. Four line pattern wheels are shown permanently mounted on the pattern wheel shaft. Selection is made by sliding the line catch to engage with the required pattern wheel. The selection is secured by a screw which fixes the position of the line catch (166 plan). (The pivot shaft for the line catch obscure the pattern wheels which should be shown dotted in the plan view in 166. The pattern wheels are, however, depicted with solid lines. This is one of several examples of casual drafting convention.)

When the line catch releases the pattern wheel there is a danger of the pattern wheel running away under the influence of the suspended weight As drawn, the pattern wheel would advance one tooth for each release if and only if the catch returns before a second tooth passes i.e. the correct operation of the drive relies on the machine being driven at the correct uniform speed to ensure that the catch returns in time to prevent the passage of more than one tooth. This is an unnecessary constraint and one difficult to guarantee. There is also the risk of runaway if the machine is halted with the line catch withdrawn.
It was regarded as inadvisable to rely on the constancy and continuity of the engine speed for correct line-to-line operation and it is unlikely that this was intended given the lengths gone to elsewhere to secure particular functions. The omission of a means of ensuring single-increment advance of the line pattern wheel was taken as a design omission requiring remedial action (see below).

End of Column Action

Stereotyping progresses down the page with successive releases of the line-to-line pattern wheel. The line rocker pivot which is integral with the oscillating bar moves to and fro carrying with it the line rocker 1B. The lower bar has a short section widened to accommodate a through-slot and the line rocker lever passes through the lower bar to engage a slot in the upper bar. On the outward stroke of the lower bar (right to left in the end elevation 166) the pad of the rocker lever is pressed against the right hand wall of the through-slot and drives the upper bar to follow the outward stroke. During the outward stroke the rocker lever is trapped at the angle drawn, and upper and lower bars and rocker assembly moves as a whole 13/16" to the left. On the return stroke the lower bar moves to the right. The top end of the line rocker lever is held by the upper bar and the line rocker turns anti-clockwise on the moving pivot leaving the upper bar stationary. This continues until the right hand foot of the line rocker meets a fixed end stop which prevents the foot travelling further to the right. (The end stop is shown as a zigzag series of right angles in the plan view of 166.) With the foot of the line rocker obstructed the line rocker lever drives the upper bar to the right as the lower bar returns and the line rocker returns to the inclined position shown. The overall effect is that the two bars execute the outward stroke together with the upper bar lagging the lower bar during the return stroke. The length of the strokes is the same (13/16") for both bars.

Integral with the pattern wheel is a line pattern arm (3D) shown clearly in 166 plan. The action described continues with the pattern arm stepping its way clockwise with the pattern wheel until the pattern arm approaches the nine o'clock position. The line rocker is displaced to the left when the pattern wheel is next released and the lug of the pattern arm fouls the foot of the line rocker. This holds the pattern arm just off the nine o'clock position. On the return stroke of the lower bar the foot of the rocker arm clears the lug and the pattern arm advances to nine o'clock with the foot of the rocker lever now trapped between the fixed stop to the right and the lug of the pattern arm to the left as shown on 166 plan and elevation. This last pattern wheel movement positions the matrix pan to receive the last stereotyped line of the column. With the foot of the line rocker trapped in its right-most position the next outward stroke of the lower bar drives the upper bar beyond the 13/16" travel of its own stroke by the lever action of the line rocker turning on the moving pivot. The additional throw of the upper bar has two effects: (1) the ramp on the upper bar (166 plan) operates the clutch lever, 5B2, which engages the clutch to rewind the line pattern wheel and thereby reposition the matrix pan at top of page; (2) the fixed pin on the left of the upper bar (as seen in end view 166) releases the catch for the column pattern wheel which drives the carriage across the page to the new column position. These two actions position the matrix pan to receive the next stereotyped result as the first entry of the new column.

The column pattern wheel (6N) has two or three teeth depending on the number of columns required, and their distribution on the outer edge of the wheel is determined by the margin widths. In the case of multiple column stereotyping the extended throw of the upper bar that occurs at the end of a column will release the column pattern wheel to rotate one tooth pitch anti-clockwise as a result of the drive provided by the falling weight (A). Again there is no provision to prevent runaway in the event of incorrect engine speed or the engine cycle halting with the column catch disengaged. The rotation of the column pattern wheel drives the pan carriage laterally (right to left facing the engine) to position the pans for stereotyping the next column of results. The drive train from the pattern wheel to the carriage is via bevel gears shown in 163 and in outline in 147. The column pattern wheel directly drives the shaft 6H (166 plan) which drives the cross-shaft ?K via the bevel gears. The column-to-column drive train is completed by the racks and gears at each end of the frame shown in 147 End View.

The forked end of the clutch lever 5B2) drives the outer clutch bar (3T) left to right as seen on Elevation 166. The clutch bar is keyed to he main drive shaft and rotates continuously during a calculation run. As the bar sweeps round, two lugs on the bar engage with two lugs on the stationary gear wheel T (166 Elevation). The lugs are 180o displaced and on different pitch circle radii so that the outer and inner lugs only engage with each other (two elevations of 166). The clutch bar drives the wheel T anticlockwise (166 end view) which rewinds the line pattern wheel via gear 4W. Only one pattern wheel is automatically rewound: 4W engages with either the line-to-line pattern wheel or with the column-to-column pattern wheel but not both. The selection is made by sliding 4W along its shaft and dropping a latch into one of two slots in the spacing sleeve. The extra thickness of gearwheel T (Elevation 166) ensures that 4W remains in engagement in either position. In the worst case of three-column format (small type only) the column pattern wheel has sufficient rotational reserve to drive the carriage across the page without rewinding. In the case of line-to-line stereotyping in multiple columns, the column pattern wheel is rewound by hand. In the case of column-to-column stereotyping, the column pattern wheel is rewound automatically and the line-to-line pattern is rewound by hand at the end of page.

The clutch is thrown out of engagement after about 340o rotation by two fixed ramps 6a, 6c engaging with mating ramps on the clutch bar. The ramp on the upper oscillating bar, which originally activated the clutch lever, has long since withdrawn and the clutch lever is driven back to the disengaged position. The line pattern wheel is held in the rewound position, with the weight raised, by the line catch which rides over the teeth during rewind, with a leaf spring providing the return force. The line pattern wheel rotates less than one full revolution in driving the matrix pans from top of page to bottom. When the pattern arm is at the start position i.e. when the pan is at top of page, the arm is therefore clear of the nine o'clock position and spurious end-of-column action is avoided during a page run since the arm only reaches nine o'clock at the end of a column. The clutch bar rotates continuously during calculation i.e. its motion is not interrupted by engagement or disengagement.

The column pattern wheel pulley has sufficient reserve of cord to drive the carriage fully across a page without rewinding; the line pattern wheel pulley has sufficient cord reserve to drive the carriage down the page once without rewind. In single or multiple columns in line-to-line or column-to-column format a full page of results can be produced without interruption for manual rewinding. In the case of single-column stereotyping no manual rewind is required at all even at end of page: the line pattern wheel is rewound automatically and the column pattern wheel is never released to rotate since no left-to-right traverse across the page is needed. Any column pattern wheel can therefore be used for single-column operation as the pattern arm rather than the pattern wheel is the effective component. Once set at three o'clock at the start of page, the column pattern arm remains in this position to trigger the end-of-page sequence. For column-to-column stereotyping, the column pattern wheel is rewound automatically at the end of each ??linnd?? the line pattern wheel is rewound by hand at the end of a page. Both manual rewinds occur with the engine halted after a page is complete.

End of Page

If the page is incomplete when stereotyping reaches the end of a column, the control mechanism traverses the carriage to the next column and rewinds the carriage to top of page as described. If the page is complete at the end of column then the control mechanism halts the machine to allow the pans to be replaced. The end of column condition is indicated by the line pattern arm reaching the three-o'clock position.
This initiates the release of the column catch as described. In the case of multiple column stereotyping the column pattern wheel increments to the next tooth position and the pattern arm rotates towards the three-o'clock position; in the case of single column stereotyping the column pattern arm is set at nine o'clock at the start of page and remains stationary throughout The end of page condition occurs when the line pattern arm is at nine o'clock signalling end of column and the column pattern arm is at three o'clock signalling the last column. The co-incidence of these two conditions initiates the end of page action via the cluster of levers and catches shown between the two pattern wheels on 166 end view.

The end-of-page action triggers the release of weight C (166 End View) suspended by a cord over stop pulley 3L. The engine is halted by when the weight is released into the trough below (166). The falling weight operates the trip lever which primes the scoop cam to disengage the main drive clutch. The clutch breaks the drive between the operator's handle and the bevel gear of the earn drive shaft (see Section D).

The stop pulley is controlled by two radial levers which are integral with the boss of the stop pulley (3L2, 3L2). These are shown between nine and ten o'clock and between ten and eleven o'clock in 166 End View. One lever is the column stop catch, the other is the line stop catch. Each of the levers is obstructed from clockwise motion by a trip lever (T1, 2T1) which rocks on a fixed pivot. Both trips need to be released to release the stop pulley: releasing one and not the other has no effect i.e. the arrangement is a mechanical AND gate requiring both release conditions to be met The line and column trips are operated by extra arms integral with the line and column rockers respectively. The arm on the line rocker is shown at roughly nine o'clock (166 End View) and that on the column rocker at three o'clock.

The end of column condition occurs when the line pattern arm reaches nine o'clock; the end of page condition occurs when the line pattern arm reaches nine o'clock and the column pattern arm reaches three o'clock i.e. the end-of-page condition occurs when the end-of column and last-column conditions are both met at the same time. If the line pattern arm is at nine o'clock the line trip is operated during the outward stroke of the lower bar by driving the line rocker anticlockwise. The tilt of the line rocker releases the line stop catch. Similarly, if the column pattern arm is at three o'clock the column rocker operates the column trip to release the column stop catch. Only if both these conditions are satisfied is the stop pulley released. If either pattern arm is not in its terminal position (for example, end of first column in a multiple column format) then only the opposite trip is operated with no net effect on the stop pulley.

The trips operate singly when one or the other, but not both, of the end-of-column and last-column conditions are met The trips also operate singly, and therefore without effect, during the return stroke of the lower bar when neither terminal condition applies i.e. while intermediate results are being stereotyped. It was observed earlier that while the upper and lower bars execute the outward stroke together, the upper bar lags the lower bar during the return stroke. During the interval of hysteresis the line rocker (in the line-to-line configuration shown in 166) tilts anticlockwise as the lower bar executes its return stroke and the upper bar remains stationary. This tilt is sufficient for the line rocker to release the line trip. The column trip is unaffected and the single release, which occurs with each oscillation of the bar, does not release the stop pulley. In the column-to-column configuration (see below) the column trip is operated each time the lower bar executes the return stroke again with no net effect.

Conversion for column-to-column Stereotyping

The configuration of the mechanism shown in 166 is for line-to-tine printing in single or multiple columns. In this format each result is impressed below its predecessor to generate an output column. At the end of a full column, wind-back to the top-of-page occurs to restart a new page in the case of single-column operation, or to restart a new column at the top of the same page in the case of multiple column operation. For column-to-column stereotyping the carriage needs to traverse right to left (facing the engine) to accept successive results on the same line, then advance the carriage to the next line, and finally wind the carriage back left to right to start a new line. For two-column stereotyping the carriage is driven left and right across the page alternately and line increments occur after each pair of results; for three column stereotyping (option with small type only) the carriage traverses twice to the left, advances front-to-back by one line increment every third result, and rewinds fully to the right for start of new line.

The conversion from line-to-line column-to-column operation is made by removing the upper of the two oscillating bars and refitting it in an inverted position. Four pairs of vertical guides, integral with the lower bar, constrain the upper bar in the horizontal plane. The upper bar is freed by removing the keeper fitted across one pair of guides shown second from the right in 166 end view. Once the bar is inverted the keeper is replaced. The two lugs on the upper bar 9v. which hold captive the upper end of the line rocker in the line-to-line configuration, face upwards in the inverted position, and are inactive. Two corresponding lugs to the left (166 end view) on the upper bar become active by trapping the end of the column rocker instead. The conversion is completed by transferring the removable pin on the lower bar (166) to a corresponding position on the left of the bar so as to operate the column catch instead of the line catch. In line-to-line mode the fixed pin at the right hand end of the upper bar faces into the page and is inactive. In the inverted position this pin faces out of the page and operates the line catch. Finally, the rewind wheel, 4W, is repositioned and fixed with the drop catch so as to disengage from the line pattern wheel and drive the column pattern wheel via idler 2M. instead.

In the column-to-column configuration the column catch is released each cycle on the outward stroke of the lower bar. The release of the column pattern wheel advances the carriage to the next column via the drive train consisting of the shaft 6H (166 plan), bevel gears (1L, 6L. ) (163), cross shaft 1K (147 End View), and gears (1N, 1H) and racks (2M1, 2M2) (147 End View). The upper bar, driven now by the column rocker rather than the line rocker, follows the lower bar on the outward stroke and the 13/16" travel is insufficient to operate the line catch. On the return stroke the upper bar lags the lower bar as before. The column rocker operates the column trip during the return stroke due to the tilt of the rocker arm during the hysteresis gap but unless the end-of-column condition is met (line pattern arm at nine o'clock) this has no effect

In the case of two-column stereotyping the column pattern arm reaches the three o'clock position after only one release and the arm will therefore reach the three o'clock position every second cycle. As the arm is driven anticlockwise by the suspended weight (A) fouls the foot of the column rocker and the arm is held just off the three o'clock position. On the return stroke the foot of the rocker clears the pattern arm and the arm nudges home to three o'clock. The carriage is now positioned to receive an impressed result in the second column. On the next outward stroke the pattern arm obstructs the foot of the column rocker and the rocker levers the upper bar to the left (166) in excess of the standard 13/16" travel. The extended throw of the upper bar allows the fixed pin to in the upper bar to operate the line catch and the release of the line pattern wheel advances the carriage by one line increment. The extended throw also allows the ramp on the upper bar (166 plan) to operate the clutch lever (5B2, ) and engage the clutch which rewinds the column pattern wheel as described earlier.

The sequence described is repeated with the a line increment occurring after alternate pattern wheel releases. The line pattern wheel advances clockwise until the line pattern arm reaches nine o'clock. The end-of-page condition is met when both pattern arms are horizontal and facing each other. On the next outward stroke the two trips are released by the rockers. This releases the two stop catches and the stop pulley releases the weight to halt the calculation run. For column-to-column stereotyping the column pattern wheel is rewound automatically, and the line pattern wheel and stop pulley are rewound by hand as part of the end-of-page procedure. For three-column stereotyping (small type only) the column pattern wheel has three teeth and the line pattern wheel is released after every two releases of the column pattern wheel.

Changing Pans

The line-to-line movement of the carriage is derived from the single drive shaft, 7A (147, all views). The line advance is driven by the central rack and pinion when stereotyping using small type, and by the rack and pinion pairs in the case of large type. It was observed earlier that the pitch of the line-to-tine advance when using small type is less than that for the large type and that since the carriage is shown as a single assembly, a given run of stereotyping would use the small matrix pan or the large matrix pan but not both. 147 End View shows the small and large pans aligned at top-of-page below the type heads with the common starting positions separated by 17.25" this a rare absolute dimension.

The type heads need to impress the soft filler and the depth of impression will require the type heads to penetrate into the body of the flong. To prevent the type heads on either side of the active heads fouling the lip of the pan at the top and bottom of the page the pan is filled so that the upper surface of the flong is flush with the lip of the pan. When stereotyping using small type the end-of-page position of the pan leaves the pan run out to the left (147 End View) with the type heads above the last line of the page; the large pan (without soft filler) will be partially advanced down the page. The automatic halting of the engine at the end-of-page and the carriage rewind that ensues prevents stereotyping overrunning the pan and the type heads are thereby prevented from fouling the right hand lip of the small pan (147 End View). In the case of stereotyping with small type, the large pan (unfilled) can therefore be present on the carriage without penalty. However, the inverse is not true: when stereotyping with large type, as the carriage traverses down the page the small stereotype heads will foul the lip of the small matrix pan as the end of the small page is overrun. The small matrix pan would therefore need to be removed from the carriage when stereotyping with large type and a spacer inserted in its stead to secure the large pan. The Timing Diagram ([F] 385/1a) shows the table clutch (column 29) engaging during interval 42 (Babbage cycle division corresponding to 302.4o) and the engine halting under automatic control at Does this mean that the events between 350o and 360o occur after restart?] following the release of the stop pulley and weight C The automatic carriage rewind commences with clutch engagement so that by the time the engine halts at the end of cycle, the carriage is already partially rewound. Rewinding continues throughout the first half of the next cycle and is completed during interval 32 (230.4o) i.e. well into the second half-cycle.

In the case of column-to-column operation, the pans will halt with the last-line position under the type heads and the carriage partially returned towards the first column (left to right facing the engine). For small type stereotyping the small pan will be run out to the left (147 End View) largely clear of the "type heads and can be removed, and if necessary renewed, with relative ease. Removal of the large pan, however, is obstructed by both sets of type heads. For large type column-to-column operation this difficulty can be overcome by reengaging the main drive clutch after automatic disengagement and advancing to the next cycle and halting the machine manually during interval 32 (230.4o). This corresponds to the completion of the automatic carriage rewind cycle and coincides with the disengagement of the table clutch. In this position the large pan is fully run out to the right (147 End View) with the type heads positioned at the start of page but with the next result not yet impressed. Removal and replacement of the pan is relatively unobstructed in this position. The automatic halting of the engine serves here as a timing marker rather than positioning the pans for removal.

In the case of line-to-line operation the carriage will be partially rewound back down the page (front-to-back facing the engine) when the machine halts automatically. In line-to-line mode the removal and replacement of small or large pans will be partially obstructed by the type heads. For small pan operation the solution is to remove the spacers and wedges positioning the pan and to move the pan towards the gap between the type heads (left to right 147 End View). This will allow it to be slid out For large pan operation the solution is as described for column-to-column operation i.e. to reengage the main drive clutch and advance the cycle until the carriage is rewound to top-of-page, the position shown in 147 End View.



The line and column pattern wheels are driven by weights, B and A respectively (166 end view) suspended over pulleys 3G, 6G fixed to the pattern wheel shafts. When the line or column catch is released by the action of the pegs on the oscillating bar the line and column pattern wheels a free to rotate driven by the appropriate suspended weight. However, unless the catches are is reengaged in time to allow only one pattern wheel tooth to pass (this is difficult if not impossible to ensure as the period of disengagement depends on the engine speed which, since manually driven, will be variable) there is the danger of the pattern wheels running under the influence of the falling weights.

Column 24 of the Timing Diagram (([F]385/la) indicates the active period of line and column catch disengagement and the catches are referred to as 'Detent 6C2 or 5C2' at the head of the column. The function of the catches is to allow the pattern wheels to advance one tooth at a time in one direction, and be overridden during rewind in the other. Their function is therefore somewhere between an escapement and a ratchet and referring to the catches as detents, which usually implies that they can be overridden in either direction, is curious.

A modification is required to provide the escapement action i.e. to restrict the pattern wheel advance to one tooth increment only for each release of the line catch. The solution chosen is to retain the first line catch unmodified from the original and to provide a second line catch to prevent unwanted advance after each release. The modified arrangement is shown on GA L24. The pattern wheel and two line catches act as a ratchet and double pawl. Both line catches must act on the same tooth as the pattern wheel teeth are not continuous but are bunched in interrupted groups of five corresponding to five-line groups of print separated by blank rows for ease of reading. An additional peg in the lower oscillating bar is provided to operate the second line catch. During the rest period the second line catch is held out The catch is released by the peg during the outward and is driven into engagement by the leaf spring: the second catch engages before the first catch is withdrawn. During the return stroke, the first catch engages before the second catch is withdrawn.

The guides for the oscillating bars require modification to accommodate the additional pegs. The original drawings show the lower bar supported at each end by channels in the two end supports (166 Elevation and end view), and the upper bar supported by four sets of guides {E) attached to the lower bar.]. To allow space for the extra pegs in the lower bar, the outer two sets of guides attached to the lower bar are combined with the end supports to provide support for the upper and lower bars (L471, L472); the central two remaining guides remain unchanged.

The modification of the column catch is more extreme though the principle of providing a second pawl is identical. The column pattern wheel assembly is crowded more closely to the end framing pieces than the corresponding line pattern wheel assembly and there is insufficient space to accommodate a second column catch without modification to the original catch. The modified arrangement is shown in L24. The original transmission is changed from direct lever action, to a lever-driven gear. This converts the operation of the catch from leading to trailing action as well as creating sufficient room for the second catch. The conversion to trailing action has the additional elegance of restoring symmetry with the modified line catch arrangement opposite. The pivot position of the peg-operated arm of the first catch is unaltered and additional pivots are provided for the pawl lever and second catch. The second catch is also geared and is driven by additional pegs in the lower oscillating bar. The phasing of the two catches is identical to that of the line catches i.e. the second line catch is released for engagement before the first catch is withdrawn, and withdrawn after the first catch reengages.

The modification have minor consequences for the pawl profile and pattern wheel teeth. The rise of the tooth on the original line catch is shown hooked for ratchet action, and a sawtooth on the reverse side so that the catch can be overridden during rewind (166 End View); the original column catch has a different tooth profile (166 End View). The double pawl action makes the hooked tooth unnecessary and the fact of the modified column and line catches both being trailing levers allows all four pawls (line and column) have radial tooth rises on the active faces. Standardising the pawl profile also means that all pattern wheels (line and column) can have the same tooth profiles; the original arrangement of one leading and one trailing lever required different tooth profiles.

Split Carriage

The two elevations in 147 show a single platform bearing the two matrix pans mounted on inverted v-shaped runners. The carriage bearing the two pans tracks on the long runners to provide line-to-tine motion (front-to-back motion facing the engine) (147 end and front elevations); the long runners track on transverse runners to provide column-to-column movement (right to left facing the engine). The drive for line-to-line motion is provided from shaft 7A via two rack and pinion arrangements. The larger line-pitch required for the larger typeface is provided by the pair of racks 1R1 and 1R2 fixed to the underside of the matrix pan platform engaging with pinions 5S and 5P respectively; the reduced line pitch for the smaller typeface is provided by the single central rack 4R and pinion 5R (working point 5a).

While the general intention seems fairly clear-there are mystifying aspects of the detail. 147 Plan and Elevation show an uninterrupted sleeve 4B on the drive shaft. The sleeve is shown keyed to the shaft by a sliding key along the full length of the shaft with the keyway extending along the full transverse travel of the carriage - this so as to provide line-to-tine drive while the carriage traverses column to column. The End View shows a key cut into the shaft. It is unclear (depending on where the section is considered to be taken) whether the pinions 5S and 5P, or the sleeve, are keyed to the shaft. If the pinions are keyed to the shaft then the sleeve acts as a spacer separating the pinions which are sandwiched by collars against the sleeve. On the other hand if the sleeve is keyed to the shaft then the pinions are integral with the sleeve and what would otherwise be outer collars become bosses which space the pinions from the frame.

The central single pinion 5R (with working point 5a) is driven from the same shaft 7A but the method of coupling between the shaft and pinion is not clear. (The fact that the central pinion is driven by 7A is confirmed by the notation of trains in ??Spon?? between pages 251 and 252, 'Trains of the Printing Part of the Analytical Engine', bottom right). Dotted lines (Plan and Elevation) indicate that the pinion fits over the sleeve though no pinning or other means of coupling the pinion to the sleeve is shown. Similarly, no means of fixing or securing the single small rack is shown: this is drawn levitating freely in 147 end elevation.

The fact that the sleeve 4B has a different index of identity to the pinion (5R) indicates that they are separate pieces. However, the notation of trains indicates that shaft 7A and a component 5B loose in relation to each other (indicated by parentheses), are driven together i.e. the 'loose' relationship is the linear transverse motion, and the driven relationship is the rotational drive transmitted by the key. If the identity index of B in 147 is taken to be an error i.e. the sleeve is taken to be 5B instead of 4B then the pinion and boss are integral with the sleeve and this resolves the problem of the transmission of drive.

The use of dotted lines elsewhere in the drawings is not always consistent If the dotted line showing the sleeve as continuous inside the central pinion is taken to be an error then the sleeve can be regarded as broken i.e. is in fact two sleeves acting as spacers to locate the central pinion in the correct position and to space off the two larger pinions. The central pinion would then be keyed directly to the shaft. It is unlikely that the sleeve is continuous and passes through the central pinion and boss: there is little wall thickness remaining (147 Plan and Elevation).

The pans are located left-to-right between the sidewalls of the carriage. The top-of-page end of the small pan, and bottom-of-page end of the large pan, register against the end-walls. No means of securing the matrix pans to the platform is shown and no means is shown of ensuring that the pans remain butted against the end-walls. If both pans are assumed to be fixed to the platform then the separate rack and pinion drives cannot be engaged at the same time since the line pitch increments for the larger and smaller types are different and the different rates of feed would result in a destructive contention between the two drives. When stereotyping using the large typeface, the small rack and pinion drive needs to be disengaged, and vice versa for small typeface stereotyping. No provision is made in the original drawings for the disengagement of the large racks and the lack of detail in the means of fixing the small rack, and the lack of indication of how the small pinion is driven leaves open the question of whether the small rack and pinion were intended to be capable of being disabled. Access to the underside of the carriage to disengage either of the drives is severely restricted and it is difficult to see how a convenient means of disengaging the drives could be devised. The function of the flanges on each side of the central pinion is not clear. They may be intended to act as guides if the rack is to be uncoupled and refitted. This notion gives tentative support to the supposition that only one of the two sets of racks are operative at any time. But the case for the flanges is marginal and their inclusion remains mystifying. If the method of uncoupling the drives involves disengaging the racks from the pinions 'it is essential that they are replaced in exactly the same mesh so as to ensure correct positioning under the stereotype heads as the pinions are phase-locked to the pattern wheel drive. If the racks are incorrectly remeshed the results would be displaced on the page; there is also the danger of the stereotype heads fouling the walls of the pans. The need to refit the racks exactly places an operational constraint on any means devised for uncoupling the drives.

The mutual exclusion of the two drives was first assumed to imply that simultaneous stereotyping using both large and small types was not intended. This conclusion was supported by the fact that if the small pan is present in the carriage when stereotyping to the large pan then the small stereotype heads would foul the trailing wall of the small pan (at the end-of-page end) as the carriage advanced down the page. The small pan could not therefore be present during large-type stereotyping without risk of damage. The difficulty of devising a means of uncoupling the drives and the operational difficulties associated with disabling one or another of the drives to transfer between large and small stereotyping led to splitting the carriage into two with one section for the small pan and one section for the large pan. Both drives remain permanently engaged and simultaneous stereotyping using both pans becomes possible. The rate of feed for the two pans are different and, as the stereotyping progresses, the large pan gains on the small pan. Because of the different rates of feed the front lip of the small pan does not pass under stereotype heads and there is no danger of fouling. A rare dimension in 147 End View shows the distance between the first lines of the small and large pans as 17-1/4 inches. Working back from the diameter of the pinions gives the maximum travel of the large pan as 14.75 inches, and 5.9 inches for the small pan i.e. a feed ratio of 5:2. Retaining the original pan sizes, pan separation at start, and pinion diameters, the pan separation at end of page is 0.1375 inches (see GA 337 L 25) - this from the common starting point determined by the position of the stereotype heads.

The decision to split the carriage was made in the light of the lack of appropriate detail in the drawings and the perceived difficulties in devising a convenient way of uncoupling one or another of the line-to-line drives. Subsequent examination of the notation lends some support to the notion that both pans were intended to operate simultaneously (i.e. that both drives were to be permanently engaged) and that the incompatibility of the line-to-tine feed rates is resolved by the small pan, driven by the central rack. being free to slide on the carriage as the carriage (and large pan) are driven by the two outer racks. The decoding of the notation remains incomplete. However, the identifier for the central rack, 4R as well as the identifier for the small pan 7K, are underscored three times while the large-pan rack identifiers are underscored only twice (147 Plan and End View). This might indicate that the small rack has an additional degree of linear freedom in relation to the larger rack. The ??Spon?? train notation of 1856 may well be significant here (bottom right). The notation indicates that the carriage H is free to move in relation to sliders G, and G is free to move in relation to frame F (H(G(F)G)H). The line-to-line motion of carriage H (and the large pan) is shown derived from racks 1R. (The pans are also show acted on by the large stereotype sectors). The column-to-column motion of H is imparted by G driven by racks 2U. However, the small pan 7K is shown as a driven piece by one (or both) of H and G as well as by rack 4R i.e. the line-to-line motion of the small pan is derived from rack 4R and not from carriage H, while the column-to-column motion of the small pan is transmitted by carriage H through working points u. (Working points u are shown clearly in 147 End View.) This lends support to the possibility that the small pan is free in the carriage in the line-to-tine direction and moves in the carriage relative to the large pan during stereotyping. The fact that no fixings are shown to attach the small pan to the carriage may be considered supporting evidence for independent line-to-tine-motion. The lack of pan fixings (large and small) could be an omission of detail in both cases. Alternatively, it could indicate that no fixings are needed with the large pan being driven line-to-line by the front wall (i.e. at the front of the engine) of the carriage, and the small pan by the central rack with pan free to slide in the carriage. In both cases the column-to-column motion is transmitted by the side walls of the carriage.

In the final design the principle of using both the small and large pans moving at different rates is retained. However, instead of having the small pan sliding within the carriage, the carriage was split into two with the separate sections of the carriage driven at the same time by their ??reseptive?? rack and pinion drives. There are two main considerations favouring this solution: the rack providing the motion for the small pan can be fixed permanently to the underside of the carriage so resolving the operational difficulties associated with uncoupling and reengaging the small carriage with its drive each time the pan needs renewal; secondly the carriage is longer than the pan alone and is therefore more stable on the v-slides slides.

The general arrangement of the new carriage is shown in GA 337 L25. Detail of the small pan rack is given in L352 (called 'rear' i.e. away from the front of the machine), and that of the large pan racks in L351 A (LH) and L351 B (RH). The rear pan is some distance away from the rack that drive it and the gap needs to be bridged. The rack is fixed to a ski-shaped support ribbed for stiffness in view of the long overhang. The top bearing surface of the support slides on the underside of the front carriage which acts as a guide. The rear carriage was extended to use more of the slide than shown for the single carriage. This provides a more stable guide for the rear carriage especially in view of the long overhang of the rack support. The new front and rear carriages are of equal size i.e. both carriages can accommodate the large matrix pan. Using small type and the large pan, reversing the pan at the end of page allows a maximum of two sets of three columns back-to-back in one pan.

Removal and replacement of the small pan is best carried out when the carriage reaches the end-of-page position i.e. when the engine stops under automatic control. To remove the large pan the engine cycle needs to be advanced until the carriage is wound back to the top-of-page position i.e. with the carriage at the extremity of its travel towards the front of the engine.

The sleeve 4B shown as continuous on the shaft in 147 Plan was split into two. The two sections locate the central pinion by sandwiching it between them. Bosses were added to the large pinions and two outer collars are fixed to the shaft between the pinions and the side members of the frame. The full-length key was dispensed with and replaced instead by three short sections of key, one for each pinion, trapped between the two spacers in the case of the central pinion and between the spacers and the outer collars in the case of the pinion pair. The sleeve is not keyed to the shaft. The use of short sections of key is intended to reduce drag as the carriage traverses column-to-column.

Column-to-Column Dashpot [Ref: Red Book. page 130]

Babbage gives no details of the pattern wheel tooth layout or of the intended options for formatting the results on a page. In establishing appropriate layout combinations (see X23) taking account of the pan sizes and font heights the maximum line-to-line carriage movement emerges as 0.275 inches. This is the maximum distance the carriage will advance in one complete calculating cycle. The falling weight that drives this motion (Weight B in 166 End View) needs to be large enough to overcome initial static friction without hesitation. If the weight is heavy enough to do this then the carriage will accelerate rapidly until stopped by the next pattern wheel tooth under direct uncushioned impact With only 0.275 inches linear travel between stops the impact and jarring should be within acceptable limits. However, in the case of column-to-column motion the carriage will step as much 5-1/4 inches between stops. The action of the second safety pawl halts the initial rotation after 1/16 inches which still leaves a full 5-3/16 inches between stops. The uncushioned impact of a heavy carriage (now made heavier by splitting into two equal carriages) accelerating over approximately 5 inches was regarded as potentially damaging and a removable oil-filled dashpot was included.

The dashpot is fitted under the carriage along the centre line of the carriage frame F (147 Plan) in direct line with the boss of the large idler gear shown in 163 between the right hand side of the stereotype table frame and the left hand vertical frame piece of the calculating section. Detail of the dashpot assembly is given in L22. The dashpot acts as a linear speed governor. Its action is unidirectional. In the damped direction (right to left) the piston forces oil through an annular aperture. During the return stroke (left to right) the piston is driven against a spring and opens the oilfeed aperture to allow undamped movement. The rate of feed for a given grade of oil is adjusted by exchanging packing washers from one side of the conical plunger to the far end of the return spring so as to alter the size of the annular aperture. The arrangement ensures that the length of the return spring remains constant for different feed rate settings. Since the carriage is wound back during the return stroke the travel is controlled and damping was not considered necessary.

Additional Fixing of the Stereotype Table

The frame of the stereotype table is shown standing on rails which run the full length of the engine. Other than the common rails, no fixing is shown between the table assembly and the calculating section framing and this was considered risky. Additional fixing was provided by extending the boss of the idler wheel (163) to pass through a bar fixed between the two vertical framing members of the calculating section. The boss is secured on the far side of the bar by a collar. This arrangement allows the correct registration of the table to avoid front-to-back tilt. Jacking points on the table legs allow for adjustment and for correction of any tilt in both planes (front/back, left/right).

Drop-height for Drive Weights

Working back to the pattern wheel pulleys from the carriage shows that a pulley rotation of 340 is required for end-to-end travel line-to-line. The corresponding length of the drop for weights A and B for a 9 inch diameter pulley as indicated (166 end elevation) is 26.7 inches (340/360 x 9 n ). However the drop from the underside of the weights to the floor is only about 11 inches. The extra drop is provided by additional pulleys suitably raised (L26). The pulleys are mounted on extension pieces fixed to the existing framework. Each of the masses A and B consist often separate weights. Each of the ten components is 'C'-shaped i.e. a section of the ring missing so that they can be stacked and removed with a no dismantling. The most appropriate mass can be determined by trial to take account of friction, traction, load and eventually wear. Some form of flexible adjustment is required to cater for the presence or absence of one or another of the matrix pans.

No means of fixing the pans to the carriage is shown whether by screw fitting or wedges (147 End View and Elevation). The solution was to retain the left-to-right fixing as a slide fit of the pan into the machined tray in the carriage and to provide screw clamps to locate and fix the trays front-to-back (L25). Both trays have two screw clamps bearing on two wedge inserts. In the case of the small pan, two extension rods bridge the gap between the rear carriage wall and the edge of the pan (rods L363, and cross-sliding guide L354). The rods and sliding guides are removable loose pieces and the clamps are tightened with knurled nuts.

Other Modifications/Additions

Pattern wheel arms

166 (end view and Plan) shows a single lug at the end of each of the pattern wheel arms, which obstructs the foot of the line and column rockers when at 9-o'clock and 3-o'clock respectively. The lugs were modified to prevent the pattern wheel arms from advancing beyond 9- or 3-o'clock. The reasoning is fairly subtle and relates to the timing of the line catch release. In line-to-line mode when the line pattern arm reaches 9-o'clock (the position for printing the last line at end-of-page) the line catch is engaged with the last pattern wheel tooth. During the second half of the next cycle the last line is stereotyped with the pattern wheel arm still in the 9-o'clock position. On the next outward travel of the lower oscillating bar the line catch is released by a peg and the pattern wheel is held only by the pressure of the pattern arm on the foot of the line rocker. If this is insufficient to hold the wheel at 9-o'clock (as it might be even when the engine is operated without interruption), or if the engine is halted at this point by interrupting the rotation of the drive handle, then the wheel will advance a small distance until stopped by the safety pawl added to prevent runaway.

The line catch now rests on the untoothed portion of the wheel and cannot reengage. The ramp on the oscillating bar operates the forked clutch lever which engages the clutch dogs. On the return stroke the lower oscillating bar disengages the runaway pawl. All that now stops the wheel running away is the engagement of the clutch. If the clutch engages successfully then a small runaway occurs and the dogs will clash as a result of backlash as they engage. A still less desirable situation might occur if the dogs clash when only partially engaged. If the clutch fails to engage through fault then we have a full runaway.

To prevent the pattern wheel arms rotating beyond the end-of-page position or last-column position (9 and 3-o'clock respectively) the pattern arm lugs were modified to form an 'L' with the horizontal foot of the 'L' engaging with the lower face of the rocker foot in both cases of line and column rockers.

Stop Cord Pulley

166 end view shows stop weight C deflected from the vertical at two points with no detail of the means by which the cord is to be routed. A cord guide was added at the upper point of deflection (detail L27).

Stop Weight

The geometry of the stop weight is uncertain. The depletion of the weights in 166 end elevation could be taken to be diagrammatic i.e. the simple rectangles not being representative of actual shapes. The only other depletion is in 163 bottom left. The completion trip is shown here strictly as a gutter though the geometry of the stop weight could be taken as a sphere or cylinder. It was first thought that the stop weight would be lifted clear of the sides of the completion trip during rewind. It would then be likely to twist on the cord under its own weight and fail to reengage when next lowered. It was therefore assumed that the weight was intended to be a sphere and the trip a form of scoop. However, improved understanding of the action of the stop weight revealed that the weight would not be lifted clear of the gutter sides and the cylindrical geometry was retained.

Resetting Handle

The pattern wheels are conveniently exposed on the outside of the control mechanism for the stereotype table (163, 166 Plan and elevation) and these can be easily gripped for manual rewind. However, the stop pulley is well buried in the mechanism and finger access for rewinding is restricted. A handle was added to assist with the rewind. It is desirable to restrict the rotation of the stop pulley both clockwise and anticlockwise. Anticlockwise restriction would prevent the weight being inadvertently lifted above the sides of the completion trip during manual rewind, as well as ensuring that the catches and trips remain associated with the appropriate levers. Clockwise restriction is desirable to prevent the cord going slack and jumping the pulley sidewalls when the stop weight reaches the end of its travel after operating the completion trip. So a handle was provided at the outer end of the stop pulley shaft to assist rewind, and two stop pins added to restrict the handle rotation to 180 i.e. between 9- and 3- o'clock (337L24).

Format Options (Drwg. Ref: 337X230)

The original drawings give no details of formatting options either as page-layout descriptions or as pattern wheel profiles. 163 and 166 simply indicate a stack of four line pattern wheels and four column pattern wheels with no detail of pattern wheel variations. The format options (numbers of columns, margin widths, line pitch and line-group layouts) were determined from scratch and the pattern wheel tooth layout derived from these.

The known constraints are the size of the two matrix pans (147) which give the maximum length and width of the two stereotyping areas; the positions of the lines of stereotyping at top-of-page for a full page of results (147 End View); and the width of the column of print for each of the two type sizes (30 by 1/16 inches and 30 by 1/8 inches). (The number of digits in a result is not variable and is always 30).

Some format features can be inferred from these known constraints. For example, when stereotyping using both pans in multiple column format, the separation of the columns in the small pan is determined by the column width of the larger type. The width of a large-type column determines the left-to-right traverse of the carriage to position the pan for the next column. The margins between columns on the small pan are therefore wider than would be the case if the column jumps were optimised for the small type. The small pan can accommodate four columns of type but if the large pan is present during stereotyping the large stereotype heads would foul the left-hand edge of the pan at the start position. If the large pan was removed to allow four-column stereotyping then the drop weights would need to be adjusted to compensate for the lightened carriage. The decision was taken to restrict the maximum number of columns to three for small-type stereotyping. The maximum number of columns for large-pan stereotyping is limited to two by the pan width.

In the absence of more detail the following formatting features were taken as discretionary: side margins, margin widths between columns for multiple column formats, line pitch, and the numbers of lines grouped together before an additional line break (five-line groups were chosen throughout as consistent with the layout of contemporary printed tables).

The number of lines in a column or on a page, and the line pitch are chosen by selecting one of four line pattern wheels A, B, C or D. All remain permanently mounted on the end of the control mechanism. Selection is made by sliding the line catch to engage with the chosen pattern wheel (166, 3371-409). Pattern wheel D is in fact blank spare deliberately left so to provide for a format combination not catered for by the other three wheels. The three line pattern wheels make provision for thirty, fifty or sixty lines per column fifty for decimal increments, thirty and sixty for hexagessimal increments (trigonometric and astronomical tabulation) with the thirty-line format and large type used where large line spacing is required.

The arrangement of columns is determined by the four column pattern wheels, E, F, G, and H. Only one of the four is active at any time and the selection is made as before by sliding the column catch to engage with the chosen wheel. All four column wheels provide options i.e. there are no blank spares in the set of column wheels. The four wheels provide options for single, double and triple column printing, with various combinations of margin widths between columns. The specification of the pattern wheel layouts is shown in 337L409 A-D, and 337L408 E-H. (The hub for the pattern wheels is shown on L41 1).

Since automatic halting of the engine and automatic rewind is triggered by the end-of-page condition (line pattern arm at 9- o'clock, column pattern arm at 3- o'clock), this was taken as the fixed reference on the page and worked back from. A page-run therefore always ends in the same place with the last line stereotyped for a given pan always at the foot of the page regardless of the number of lines chosen (30, 50, or 60). Mapped back to the pattern wheel this means that the last tooth of each wheel is in the same angular position and the pattern wheels are in fact dowelled through at this position (337L409, L41 1). On the line-pattern wheel the position of the first tooth varies depending on the line pitch.

The same reasoning applies to the column pattern wheels. The column start position is wound back from the pan position for the last column (right-most on the printed page) and the last column is always positioned the same distance from the right-hand vertical edge of the pan. The smaller columns are centred on the larger columns because of the fixed relationship between the large and small stereotype heads (see 173). These constant right-hand margins are 3/4 inches for large type, and 1-11/16 inches for the small. For single column working these margin constraints still apply i.e. the single column will appear biassed to the right of the pan.

If the large pan is used in the position normally occupied by the small pan then the most compressed format can be achieved by stereotyping two pages of results in one pan. This occurs if two lots of three-column results, 30, 50 or 60 lines long are arranged back-to-back. This is achieved by removing the pan after the first page run and reversing it before replacing it on the carriage to receive the next run. The resulting layout is shown in X23, bottom left). The bottom three views show examples of other layouts. The middle view, for example, shows a 2 column layout in small type withrow column margins.

The various combinations of line and column formats is summarised in the Summary Table in X23 (bottom right) with a few examples of the layouts in the diagrammatic views. The top half of X23 shows the two carriages with large and small matrix pans. The view at top right shows the large pan positioned to receive the first line of the first column. The centre view of the large pan shows the end position i.e. positioned for the last line, last column. The two views of the small pan correspond i.e. are depicted as they would be for simultaneous stereotyping with large and small type. The columns of large and small type are therefore aligned vertically with the spacing between columns determined by the width of the larger type. The gap between the carriages at the end-of-page is a constant 0.137 inches.

Format Changes

The procedure for restarting a run after automatic halting at end-of-page would be to rewind the column pattern wheel by hand (for line-to-line working) and reset it appropriately at the start position for either two or three column operation. For single column operation no column rewind is required as the pattern wheel arm remains in the 3-o'clock position throughout and any of the four column pattern wheels can be used. The position of the carriage when the engine halts under automatic control is the most convenient one at which to renew the small pan. The engine is then turned over and the carriage is wound back to top-of-page automatically. With the carriage rewound to the start position this is the most convenient time to renew the large pan as most of it is now clear of the large stereotype heads. For complete format change i.e. changing from line-to-line to column-to-column operation as well as line number and number of columns, reselect and set the new arrangement of the oscillating bars, reselect the appropriate line and column pattern wheels, and set to the start position by hand either the line pattern wheel (for column-to-column working) or column pattern wheel (for line-to-line working).

Whichever is not set by hand will be automatically rewound to start by the clutch. The selection of line or column pattern wheels for automatic rewind is made by sliding the rewind idler gear along its axis to mesh with one or the other as required.

Variable Rewind

In the original design the clutch-driven rewind returns the carriage from the end-of-page position to the top-of-page position i.e. rewind is always full-length. For fifty and sixty-line formats, which fill the page top to bottom, the full-page rewind is unproblematic. However, a potential problem arises for thirty-line working. Only about half the circumference of the thirty-line pattern wheel will have teeth (194 from first to last tooth, 377L409) with the remaining sector blank. Since a run always finishes with the carriage at the limits of its travel (back-to-front facing the engine), it is the start position of the run that varies and the thirty-line run will begin about halfway down the pan. The first section of the pattern wheel will therefore be untoothed. The problem is that when the clutch dogs disengage at the end of rewinding to top-of-page, the carriage will run away until the line catch engages with the first tooth. A runaway of nearly half the full travel of the carriage would cause a damaging crash when the line catch finally engaged, and this should be avoided.

The thirty-line option is not specifically indicated in the original design and it could be argued that the runaway problem is a consequence of arbitrarily choosing a reduced page length. However, even if the original design was restricted to full-length formats the effects of fixed rewind still affects operation in column-to-column mode. Once again the last column position is the fixed reference and the start positions for two and three column formats differ. If separate column pattern wheels are dedicated for two and three column formats then a two-column pattern wheel would have roughly its first third blank and the problem of the column catch crashing into engagement on the first pattern wheel tooth persists. (The original design makes no provision for damping or governing the column-column traverse of the carriage.)

If the amount of rewind could be made adjustable then the carriage could be rewound by the clutch to an appropriate start position and the potentially damaging situations avoided (both line-to-line and column-to-column modes). An additional advantage of an adjustable rewind is that the number of formatting options using four column pattern wheels can be increased.

Instead of dedicating separate pattern wheels for two and three column working, each of the four wheels could provide for full three-column working but only part rewound for two column working. (Any of the wheels can be used for single-column working as the pattern wheel does not move.) The column pattern wheels can then be used to provide a wider range of margin width options.

It is not evident from the drawings whether the original design was intended to cater for variable rewind. The end and front elevations in 166 show two rewind lugs 2r, 2t but there is nothing to indicate whether the angular position of these lugs on the plate are variable so as to provide an adjustable amount of rewind. Reference to the drawings is inconclusive with respect to establishing the original intention of the design. To solve the problem of the start-of-page runaway in line-to-line mode in thirty-line format, and to provide the extra range of formatting options using the prescribed four column pattern wheels, provision was made to vary the amount of rewind derived from the clutch. Details of these arrangements are shown on L521 and L522.

The principle of the device is based on being able to alter the angular position of the rewind lugs in relation to the rewind gear. This is accomplished by mounting the two rewind lugs on a backplate the angular position of which can be varied in relation to the gear. The rewind gear is drilled with eighteen holes at 20o intervals and the plate is fixed to the rewind gear by two ring screws which pass through any of the pairs of diametrically opposite holes. To vary the amount of rewind, the ring screws are removed and the plate rotated until the engraved mark lines up with the pattern wheel letter on the rewind gear. Replacing the ring screws fixes the plate in the new position. (GAL27 gives a clear view of the clutch assembly). Twenty holes provide a series of fixed angular increments which reduce the start-of-page runaway to an acceptable level.

Go to next section Calculation
Go to top of Babbage Technical Description