Go to top of Babbage Technical Description

Table of Contents of this web page
- General
- Coupling to the Calculating Section
- Vertical Racks
- Implications of Rack Offset
- Framing and Horizontal Racks
Paper Printer
- Inking
- Printing Stroke
- Paper Feed
- Paper Feed - Design
- Locks
- Cams
- Figure Rack Lock
Paper Printer - continued -
- Paper Drum Lift
- Printing Sector Lock
- Vertical Rack Locks
- Stereotype Sectors
- Stereotype Sector Locks
- Ink Roller Cam
- Inking
- Drive
- Pinion Offset
- Printing - Design
- Loading Paper
- Counterbalancing Vertical Racks



Drawing References (main drawings in bold):
BAB [A] 147, 161, 162, 163, 164, 165, 166, 172, 173, 174, 175, 176

Integral to the concept of the Difference Engine is the automatic production of results by mechanical means. Babbage's intention was that the 'unerring certainty of mechanism', as Lardner put it, would eliminate the risk of human error in the process of calculation, recording results, transcription into tabular form, typesetting using loose type, and finally, proof reading --- all processes ordinarily performed manually.

The printing apparatus appears on the left hand side of the vertical framing and is directly coupled mechanically to the calculating section (163). The whole mechanism is often referred to for convenience as the printing mechanism, or printer. However, though the apparatus does produce a printed record on a paper roll. The primary function of the whole mechanism is stereotyping i.e. impressing numerical results on paper mache or soft metal. The stereotyped plates were intended to be used as moulds for making printing plates from which the tabular results would be duplicated.

The overall mechanism consists of two distinct sections: the stereotype table apparatus, and the printer and stereotyping apparatus. The stereotype table, shown as the lower section left in 163, is larger front-to-back than the calculating engine (164) and is supported by its own cast stand which arches over the two long members of the engine base. The table apparatus automatically positions the trays of paper mache, or strips of soft metal, under the print heads for each impression of a result and automatically repositions the trays at the end of each calculating cycle ready for the next impression. The layout of the results on the page is programmable by interchanging a selection of pattern wheels. A variety of column formats are available. Single column layouts are produced by impressing results on successive lines: multiple column formats are generated in a column-to-column sequence across the page or as two vertical columns wrapping round from bottom to top. Separation between lines is also programmable in fixed increments determined by the pattern wheel selected and results can be stereotyped with large and small typefaces.

Only the 30 least significant digits of the 31 digits result on the figure wheels are transferred to the printing and stereotyping apparatus: the most significant digit, represented by the uppermost figure wheel on the tabular column, is not transferred and is therefore not printed or stereotyped. This is digit is probably intended as a guard digit to act as an overflow warning.

The printer and stereotyping section, shown as the rectangular assembly above the table apparatus (163), contains sets of number wheels which are lowered once each cycle to impress the results on the soft material below. A full 30-digit result is impressed in one action i.e. all digit positions are impressed simultaneously. Babbage calls the number wheels stereotype sectors (172) as a portion of the circumference of each wheel has embedded in it ten heads embossed with the numerals 0-9 (172). The stereotype sectors have two degrees of freedom only: rotational and vertical i.e. each sector can rotate about a horizontal axis to position the correct head for a digit, and the whole assembly is lowered to impress the result The stereotype sectors are not free to move in the horizontal plane i.e. the position of the result on the page is determined by the XY motion of the stereotype table. The apparatus provides options for two type sizes and spacing of print. The two print pitches are fixed i.e. 1/8" and 1/16". The two sets of print heads act in tandem i.e both are lowered together regardless of which is actively impressing the material in the trays. Impressions from both sets of stereotype sectors can be obtained in the same run.

In addition to the stereotyping function, the printer section prints a hard-copy record of each result as part of the stereotyping cycle. The tabular result is transferred to print wheels (printing sectors) at the same time as the stereotyping sectors are set. The apparatus contains a paper roll feed, an ink bath and a system of inking rollers, and printing sectors which automatically produce an ink-on-paper printout, one result per line, on a continuous paper roll. The roll is automatically advanced each cycle. The printer is a valuable aid for testing and during setting up; it also provides a hardcopy record of all results.

The printing and stereotyping apparatus is driven by gears pinned to a shaft running the full length of the engine below the calculating section. The drive for the shaft is derived from the main cam drive shaft via a bevel gear on the underside of the cam stack (163). There is no printing overhead in the timing cycle i.e. the stereotyping and printing action does not lengthen the timing cycle but takes place in parallel with the second half of the second half-cycle.



The tabular value is transferred from the last even figure wheel column to the printing and stereotyping apparatus by a transmission train consisting in turn of the tabular figure wheels, sector wheels, horizontal racks, pinions, and horizontal spindles (174, and 176 bottom left). The spindles drive a set of vertical racks via a second set of pinions. The vertical racks project downwards into the printing and stereotyping apparatus and form the final link in the transfer from the calculating section.

The general operation of the printing and stereotyping apparatus is shown on elevation 172 and plan view 173. The lower section of the vertical racks mesh with the circular printing sectors and the downward motion of the racks transfers the thirty-digit number to the .set of printing sectors. The anticlockwise displacement from zero of the printing sectors positions the print heads in a straight line at three o'clock ready to receive the impression first of the inking roller, and subsequently of the paper roll. The trajectory of travel and impact position of the inking roller and paper roll is shown dotted in 172.

The printing sectors drive a set of horizontal racks (called stereotyping racks) which run front to back. Geared to the underside of the racks are two sets of stereotyping sectors. The right hand set have a fixed 1/8" pitch; the left hand set have a 1/16" pitch. The narrowing of the horizontal racks to achieve the reduced pitch is clearly shown in plan 173. The vertical racks drive the printing sectors, which drive the horizontal racks, which in turn advance the two sets of stereotyping sectors. The tabular result is transferred to both sets of thirty stereotyping sectors simultaneously. The motion of the racks positions the two rows of print heads facing downward at six o'clock. At the appropriate point in the cycle both sets of stereotype sectors are lowered to impress the soft material in the trays on the stereotyping table below. The stereotype sectors separate from the racks during this process and reengage on completion.

Provision is made for locking various pieces of the apparatus. V-shaped locking notches and locking bars are shown operating on the vertical racking, and the printing sectors. Locking notches are also shown between the type heads on both sets of stereotyping sectors. A claw-lock mechanism is shown for the small stereotype sectors. Despite appearances only one locking bar is active at any time: depending on displacement of the stereotype sector either the left or right locking bar engages with the v-notches between the type heads. The left hand bar does not operate on the sector gear teeth. The detail of this locking mechanism is not redrawn for the large stereotype sectors, though it is clear from the provision of locking notches between the type heads, mountings for the lock mechanism (172), as well as the dovetail slider and forked drive arm (173), that the intention is to lock the large stereotype sectors in this way. In all, the printing and stereotyping mechanism is locked at five separate points in the train. The fifth lock is the vertical bar locking sector racks (174).

The apparatus for producing a hard-copy record on a print-roll is shown on the right of 172. The cluster of rollers (top right) is an inking system. The final inking roller is lowered onto the line of print heads after the result has been set up. The apparatus below is a paper feed system which incrementally advances the paper each calculating cycle and propels the paper roll in a circular trajectory onto the inked print head to provide a continuous printed record of results.

The circular apparatus shown left centre in 172 is a side elevation of the cam stack which provides the drive and phasing for the various operations. The main drive wheel is shown inside the frame at the top of 173. This drives the main cam shaft and the cam stack consisting of fourteen cams clustered at the top of 173 as well as one additional eccentric, outside the frame at the bottom of the drawing, which drives the links to propel the ink roller.

A plan view showing the layout is given in 173. The small stereotype sector assembly is shown on the left and the large sector assembly is shown centre right. Both sets of sectors have dovetail sliders which allow the vertical punching motion required for stereotyping. The top bar of the II shape (Q) drawn over the printing sectors (R) is the locking bar which traverses the full run of aligned v-notches cut into the top of the printing sectors.

Coupling to the Calculating Section

The result on the tabular column is transferred to the vertical stack of 30 horizontal racks of the printing apparatus by an intermediate sector axis (R0, 176, 161). This last sector axis, interposed between the racks and the tabular column, is an even sector axis i.e. the reverse-handing of alternate axes leaves this last sector axis unchanged. The tabular value is given off to the even sector axes, R0 included, during even to odd addition i.e. at the start of the second half cycle. Each sector of R0 engages with a horizontal rack (176, 161). The arrangement translates the anticlockwise rotation of the sector into a linear displacement of the rack in which the travel of the rack is proportional to the tabular value.

The R0 sectors act as normal even sectors i.e. they restore the tabular value after giving off in the same way and at the same time as the other even sector axes i.e. in the half-raised position: the rotational travel of the drive arm for this last even sector is the same as for the other even sectors. The return of the sectors to the zero both restores the tabular value and returns the racks to their home positions.

Though the last even sector axis functions as a normal even axis, the R0 sectors differ from those of the other three even axes, R2, R4, and R6, in respect of depth of teeth, number of teeth and angular orientation. The opened out view at the top of 171 shows the standard calculating sectors with two runs of teeth of different depths: the run of teeth that engage with the figure wheels to the right are full depth, and those that engage with the figure wheels to the left are single depth so as to allow partial disengagement in the intermediate vertical position (see Calculating section). However, both runs of teeth on the R0 sectors are full depth. This is shown in elevation 174 top right. Here the sectors, shown fully raised, are disengaged from the tabular figure wheel but still engaged with the racks. The R0 sectors in fact never disengage from the racks. The R0 sectors have one more tooth than the standard sectors. This is clear from the 176 bottom left which shows 13 R0 sector teeth for engagement with the rack compared to 12 for sectors on other axes. Finally, the angular orientation of the rack sector teeth is different (176, 161): 161 shows the limits of travel of the R0 sectors bounded by the solid line at eight o'clock and the dotted line at 27 minutes past; the limits of travel of the other even sectors is shown as bounded by the solid line at 10 o'clock and the dotted line at 7 o'clock. The sector zero stop (2M0 is shown scalloped out to accommodate maximum displacement of the sector.

Each sector rack (called figure racks in the Timing Diagram 385a) have two sets of teeth at right angles: vertical teeth in the horizontal plane, and horizontal teeth in the horizontal plane. The horizontal teeth mesh with the R0 sectors; the vertical teeth mesh with a stack of staggered pinions. The pinions drive a set of horizontal spindles fitted with another set of pinions at the left hand ends. These mesh with a set of vertical racks shown in elevation in 174 left and in plan in 173. The overall function of the arrangement is to transfer the displacement of the tabular figure wheels into vertical displacement of the racks. The sector racks are shown in plan on 176 bottom left, and both sector racks and vertical racks are shown in elevation on 174. The pinions are staggered front-to-back and top-to-bottom i.e. they are arranged vertically in pairs with the even numbered digits (N) offset upwards and backwards (176, 174 and 172). The odd-numbered pinions engage with the racks below; the even-numbered pinions engage with the racks above (174). The odd- and even-numbered racks are different so as to accommodate the horizontal staggering and vertical pairing. (The racks are detailed in modem drawings K312, 313: horizontal staggering was dispensed with.)

The elevation in 174 shows the four lower-most tabular figure wheels (units, tens, hundreds and thousands) coupled digit-for-digit to the vertical racks. The upper section of the tabular column is not shown. Though the tabular column records a thirty-one digit result, there are only thirty vertical racks shown in elevation 174 and plan 173. Only thirty digits of the thirty-one-digit result are printed i.e. the result from the top-most tabular figure wheel in not transferred but probably used as an overflow check.

The v-shaped notches on the sector racks shown in 176 and 161 are locking notches for the vertical locking bar. The operation of the lock is shown in elevation in 174. The single locking bar serves all thirty digits. The lock is engaged by the downward motion of link arm 6H This operates drive link 4V turning on a fixed pivot, which lowers and engages the lock. No detail of the upper pivot is shown. A slave rocker, which mimics the action of the right hand half of the drive link, was added in the upper section to guide the lock and ensure that it remains vertical during the locking and unlocking action.

Vertical Racks

Each vertical rack has two short runs of teeth, one set in the upper section, and one set in the lower section. The teeth in the upper section mesh with the pinions driven by the sector rack spindles (174) and the lower teeth mesh with the printing sectors (172). The overall arrangement transfers the circular motion of the tabular figure wheels to circular motion of the printing sectors.

Though not shown it is assumed that all thirty vertical racks are run to the full height of the machine. The racks slide against each other and the whole set is sandwiched between two vertical framing side-pieces 4P and 5P (174) which act as constraining cheeks. (Framing pieces are not shown in 174.) The racks are further constrained by a guide-bar which threads through slots at the lower end of all the racks. The guide-bar is shown clearly in 174 fitted into two counter bored slots in the side pieces. The guide-bar both constrains the rack motion to the vertical and acts as a tie-rod for the assembly. 172 centre shows the guide slots in the racks and rectangular cross-section of the guide bar (U).

The nominal width of the rack pinions is 1/4" (manufacturing width is specified slightly under for clearance) and the pitch of the racks is 1/8" i.e. there are eight racks per inch. The increased width for the rack teeth is achieved by lapping adjacent racks in two mating L-shapes. Detail of the lapping appears in 173 which shows two pinions (digits one and two) and three lapped pairs (digits one through six) in cross section. The effect of lapping is to double the effective width of the gear teeth to twice the rack width. As a result of lapping the rack teeth and pinions for the thirty odd-numbered digits face the front of the engine, and those for the thirty even-numbered digits face the rear (173). Because of the mirrored arrangement of pairs of sector racks, alternate spindles rotate in opposite directions (174) i.e. the odd-numbered spindles rotate into the page and the even numbered spindles out of the page. The lapping arrangement corrects this so that the uniformly counter-clockwise displacement of the tabular figure wheels from zero results in a uniform downward displacement in any of the associated vertical racks.

In addition to lapping, alternate rack pairs are offset The intention here is for pairs of racks to act as guide channels for the pinions in between so as to reduce the risk of pinions fouling adjacent racks due to side play in the spindles. The system of offsetting is repeated below where the racks act as guide channels for the printing sectors (173). Though both upper and lower sections of the rack are offset, only the upper sections, which mesh with the pinions, are lapped. This can be seen from the cross-section view in 173.

The vertical distribution of the pinions and spindles follows the vertical pitch of the figure wheels. A length of toothed section is shown in 174 which indicates that the upper sections of the racks are geared only where needed as shown for the lower rack sections in 172. The positions of the toothed sections on the upper portions of the racks will therefore be staggered diagonally bottom right to top left across the face of the rack assembly. The vertical racks are therefore not identical. The pitch of the vertical lines in 174 is that of a single rack width (1/8"). Strict interpretation of 174 would indicate that the lapping is confined to the toothed sections only since, if the lapping extended, alternate vertical lines would be dotted. However, the use of broken lines to indicate hidden edges is not consistent elsewhere and 174 was not taken as a conclusive indicator that lapping should not be extended as guides for the sliding motion (see Implementation). (K331 A-Q gives modern detail of odd digit racks; K332 A-Q gives evens).

The upper section of the rack assembly is not shown. The guide bar assembly which constrains the lower section of the racks was not duplicated here. Rather the upper section of the racks are trapped front-to-back between the two sets of pinions which act as a restraining cage. The intention to offset alternate racks front-to-back is confirmed in 172 centre top which shows the lowermost four spindles and pinions for the first two digit positions. The spindles of the upper pair (digits three and four) are shown offset to the right. This offset helps to clarify an apparent discrepancy in the front-to-back spacing of the spindles. The separation of spindles for the first odd and first even digits is consistent in 176 bottom left, 173 centre top, and 172 centre top. 161 bottom left shows this separation as slightly wider. However, the wider spacing in 161 is consistent with the spindle positions for the second and third digits rather than the first and second digits i.e. the positions shown in 172 for the spindles bottom left and top right. [Is M and N notation generic or decade dependent?]. (The M notation indicates even-powered digits (units, hundreds, ...): the N notation indicates odd-powered digits (tens, thousands, ....)

Implications of Rack Offset

Since the printing sectors are concentric and pivot on the same shaft, offsetting the vertical racks alters the pitch circle radius of alternate sectors. Equal downward displacements of the vertical racks will therefore result in different angular displacements of alternate printing sectors. A 0.2" difference in pitch circle radius (the size of the offset) results in about a 2.5o difference between the angular displacement of a sector set to print a zero and one set to print nine. If the pitch of the type heads were identical for all the printing sectors then the nine would be positioned as much as a full character height above an adjacent zero. This difference is clearly too large to be ignored. The implication is that the pitches of the locking notches, and that of the type heads, need to be altered for alternate printing sectors to ensure that all the print heads line up on the print line and all the locking notches are aligned along the line of the common locking bar whatever numbers are registered on the sectors. Only one printing sector is shown in elevation (172) and there is no indication of any provision for non-identical sectors.

If offsetting is regarded as a desirable aid to maintain alignment between the printing sectors and vertical racks, then it is, by the same reasoning desirable for the meshing of the printing sectors with the horizontal racks. Offsetting would also be desirable for the meshing of the stereotyping sectors with the lower edge of the horizontal racks. Here both sets of stereotype sectors separate from the horizontal racks when lowered to make an impression, and it would seem that the guide channel effect of offsetting alternate racks would assist alignment during re-engagement. The implication of offsetting alternate horizontal racks is that the pitch of the type heads would need to be different to ensure that they line up at six o'clock for all the sectors whatever the number set up. As with the printing sectors, there is no evidence in 172, or elsewhere, of provision for non-identical stereotype sectors.

The issue of how far into the drive train to extend the rack offset is resolved by the geometry of the stereotype sectors. In the case of the printing sectors the run of locking notches is separate from run of type heads. However, in the case of the stereotype sectors the locking notches alternate with the type heads on the circumference of the sectors. Increasing the pitch of the type heads on alternate sectors to compensate for rack offset so that the type heads align at the six o'clock position for any combination of numbers conflicts with the alignment of the locking notches for the correct operation of the common locking bar operating at the twenty-past the hour point. The two separate conditions of a fixed print position and a fixed separate locking point cannot both be satisfied if the pitch of alternate sectors differs and the type heads alternate with the locking notches. In the case of the printing sectors the two conditions are reconciled by placing the sequence of type heads and locking notches into separate groups. In the case of the stereotype sectors there is insufficient space on the circumference of the significantly smaller stereotype sectors to accommodate separate runs of type heads and locking notches. The conclusion is that the lower sections of the horizontal racks and the stereotype sectors cannot be offset.

The upshot of this conclusion is that the offsetting alternate racks must be extended to the gearing between the printing sectors and the horizontal racks to correct for the differing angular displacements of alternate printing sectors resulting from offsetting alternate vertical racks. Offsetting the gearing between the printing sectors and the horizontal racks ensures that the horizontal displacement of all stereotyping racks is the same for equal displacements of the vertical racks. Since the pitch circle radius of the two runs of teeth of each of the printing sectors is the same for any single printing sector (though different for alternate sectors) carrying the offsetting through to this last stage results in a 1-1 relationship between vertical rack displacement and horizontal rack displacement for all thirty positions. Again there is no indication of any provision for non-identical racks or printing sectors in the original design.

Framing and Horizontal Racks

The overall structure of the printing and stereotyping sector assembly is that of a large rectangular box frame made up of cast framing members bolted together. The frame is formed at the narrow ends by two end pieces N1 and N2, and by long side pieces, M1 and M2 (173).

The horizontal racks are boxed in at the full width end by the main side piece framing members (M1, M2) At the narrower end the straight sections are sandwiched between two edge blocks. The curved sections are contained sets of keepers, four above (5Q2, 6Q2, 7Q2, 8Q2) and four below (1Q1, 2Q1, 3Q1, 4Q1) (172), which span the main framing members, M1, M2. The straight sections of the horizontal racks are in contact and slide against each other i.e. they rely on each other for correct registration. The curved sections have small tapered clearances between them and will touch during operation depending on the combination of numbers in the result. The curved sections of the racks are formed in the shapes shown and retain their curvature when relaxed and do not rely on lateral contact to retain their form i.e. the curvature is not a sprung form resulting from blocking the straight ends of flexible strips.

Paper Printer


The provision for inking the printing sectors is shown on 172 top right The general principle of operation is shown in a diagrammatic way as a cluster of rollers but the drawing gives no detail of drive or support. In this respect, the inking mechanism is less detailed that the rest of the printing and stereotyping mechanism. A small amount of additional detail is given in 173 which shows the inking roller supported between two levers (1K1, 1K2).

The elevation in 172 shows four rollers. The ink pot roller (Z) has a scraper plate bearing on the axial surface of the roller. The scraper, roller, and side cheeks (latter not shown), form a reservoir which holds the ink supply. Printing ink is typically viscous and the arrangement, standard at the time, is sufficient to prevent leakage of ink past the scraper. The ink bath is shown fitted with a hinged dust cover.

The ink-pot roller bears on a second roller (J) which in turn bears on a third roller (K) which effects the final transfer of ink to the inking roller (I). Three of the rollers are shown suspended in space with no visible means of support, and there is no detail as to the method of drive, whether intermittent or continuous, or any indication as to the composition of the roller material.

The final inking roller is propelled onto the line of type set up with the thirty-digit result. The inking roller swings on two arms 1K1, 1K2, pinned to a shaft (1K) which is supported at each end by the two outer framing members (1J2, 7J2) i.e. the shaft spans the full width of the whole apparatus. The shaft is driven by a drive arm (1K2) (173 right bottom) which is in turn driven by link P which pulls the roller onto the line of type. The arc of the roller trajectory is shown in 172.

The provision of no more than a skeleton outline of the system of inking rollers (172 top right) could arise from the assumption of well-known contemporary practice. Given the degree of detail elsewhere, it is more likely that the omission of detail is an incompleteness in a partially finished drawing. The missing detail was supplied by reference to contemporary practice (see Implementation).

Printing Stroke

The basic printing and paper feed system consists of two drums or barrels, and a print roller. The general arrangement of the mechanism is shown in elevation on 172 with a partial plan in 173. A more detailed plan section of the take-up barrel assembly is given in 165: 'Elevation', Fig. 7, and end view, Fig. 6. The 'Elevation' shows the printing mechanism in fully raised position as viewed from the front of the engine i.e. looking left to right at the 'End View' shown in Fig. 6.

The paper path is shown in 172 and 165, Fig. 6. Paper is threaded from the roll wound on drum 1F, around the print roller, F, and onto the take-up drum 1H. The rest position of the assembly is as shown in 172 and in the dotted position of 165, Fig. 6. The relationship between the axes of the two drums and the roller is fixed i.e. the two arms, 7D1 and 7D2, carry the shafts of the drums and roller, determine the relative positions. To print a result the whole mechanism turns on the fixed pivot (D7) and propels the printing roller in a circular trajectory to press against the line of type heads. The position of the assembly at the completion of the upstroke is shown as the solid view in 165, Fig. 6. During the printing stroke the whole assembly moves as a unit i.e. there is no relative motion between the drums and the roller.

The upstroke is driven from the single heart-shaped cam, 2A4, shown in elevation on 175, and in plan as part of the cam stack towards the top of 173. A train of links and arms transmits the drive to the drum assembly. The angle of the elbow of the follower arm (H) is fixed and the two fixed pivots in the chain are the follower arm pivot (4B) and the take-up spool axis, (7D) (172, 173). The two floating pivots are the forked knuckles at the end of drive link 2J. The follower arm, H, pivots on 4B and drives link J diagonally downwards, and arm 7C (173), keyed to shaft 7D levers the drum into the upstroke.

The drive is unidirectional i.e. the return stroke is not actively driven. The assembly passes top dead centre during the upstroke (165, Fig. 6). To avoid the assembly being stranded in the printing position a counter weight is provided initiate the return stroke. The counter-weight, 7C2, is shown as a circle on 172 right bottom and is fixed to a lever arm attached to the boss of 7C (173). The counter-weight acts on the shaft on which the whole mechanism pivots and the return stroke is a controlled descent restrained by the return pressure of the cam follower on the retreating profile of the heart-shaped cam (175). During the return stroke the print roller retraces the trajectory of the upstroke. The rest position at the end of the return stroke is determined by the end stop 7C3??, which bears on the vertical frame (172). The plan view (173) does not show the end stop lever but the notation indicates that it is part of the ball-weight lever assembly (7C2).

Paper Feed

Description of the paper feed requires further exploration of the assembly shown in section in 165, Fig. 7. Each of the two swinging arms, 7D1, 7D2, which support the drums and roller, are integral with a long boss. Each boss is pinned to the shaft 7D, outside the swinging arms. The bosses extend inside the arms, sleeving the shaft, and meet in the middle of the assembly. The boss on the side with the ratchet gear has a cam fixed to it (165, Fig. 7 left), the outline of which is shown in two positions (dotted and solid) in 165, Fig. 6. A second assembly is free to rotate on the shaft formed by the boss. This consists of a sleeve, the ratchet gear. and the gear wheel (Fig. 7 right). The gear wheel meshes with a similar gear wheel fixed to the feed drum. The third part of the nested assembly is the barrel of the take-up spool. This rotates on the sleeve which carries the ratchet and gear wheel. Trapped in a recessed annulus between the take up spool and the sleeve is a clock spring (Fig. 7, right) pinned at one end to the inside of the take-up barrel and the sleeve at the other. The wound clock spring biasses the take-up spool to rotate anticlockwise (172) and maintain positive tension on the paper. A feed pawl (5L) pivots on the question-mark bracket 3K ( 172) and is biassed by a leaf spring (5E) to engage with the ratchet teeth. The feed pawl is shown engaged in 172 and lifted clear by the cam in 165. The final piece of the assembly is a backstop pawl (172) which engages with the ratchet teeth under pressure from a second leaf spring. There is no provision in evidence for fixing the paper roll at either end. In the rest position the feed pawl engages with the ratchet teeth (172, 165 Fig. 6). During the printing stroke the feed pawl is lifted clear of the teeth as shown by the solid view in 165, Fig. 6. The backstop pawl holds the ratchet and gear assembly in fixed relation to the swinging arms during the upstroke. The backstop pawl ensures that the assembly moves as a unit during the printing stroke with no relative motion between the drums.

Towards the end of the return stroke the cam releases the feed pawl which engages the ratchet (dotted view 165 Fig. 6). This halts and holds the ratchet and gear wheel in the position at engagement. In returning to the rest position the remaining travel of the swinging arms drives the gear of the feed drum against the stationary gear wheel of the take-up assembly i.e. the feed drum assembly and stationary gear act as planet and sun gears. The ratchet overrides the backstop, pawl which remains in sprung contact throughout The feed drum is driven clockwise and feeds a short length of paper from the roll. The paper is taken up by the take-up spool driven by the clock spring and tension is maintained. The overall action presents a fresh area of paper on the printing roller ready for the next impression. The length of paper incrementally advanced depends on the fixed angle through which the feed drum is driven at the end of the return stroke, and the diameter of the paper roll on the feed drum. As the paper stock depletes, the length of paper fed each printing cycle, and therefore the line height of the print, will vary slightly. The direction of the variation will be to reduce line height as the calculation run progresses.

Paper Feed - Design

The design of the paper feed arrangement is of interest The unloaded diameter of the feed and take-tip drums is shown in 165 Fig. 7 as 3" (The note written in the rectangle representing the feed drum confirms that the barrel as drawn is unloaded: "Printing paper coiled on this barrel at the commencement".) The two views (165 and 172) showing the routing of the paper between the drum also show the diameters as 3" with no indication of the effect on the diameters of paper stock. The 174 plan view shows what appears to be a single thickness of paper stretched between the roller and the take-up spool. This suggests that the printout is intended for short print runs only (checking and testing) and that the printer roll would therefore be replenished each time a stereotyping tray is changed. If only a few turns of paper are loaded at time then the effects of the gradual change in the effective diameters as the reserve stock transfers from the feed roil to the take-up drum can be ignored.

The housing for the clock spring (165, Fig. 7) is relatively small (the height of the channel is about 0.5") and the modest number of turns of the take-up drum available to pre-tension the spring would seem to limit the length of paper the spool can take-up. A further concern is that the tension of the clock spring may override the backstop pawl and cause forward runaway. The feed pawl is little use here and is in any event lifted clear of the teeth during the upstroke and the backstop pawl offers the only resistance to runaway during this part of the cycle.

The maximum capacity of the printer is worth further exploration as it exposed a subtlety in the design that was not at first obvious. The maximum paper stock possible is physically limited by the gap between the feed drum and take-up spool. This gap is about 0.58" (165, Fig. 7)- With paper thickness of 0.004" the number of turns is approximately 145 and the approximate length of the reserve paper roll is at least 113 feet (assuming a constant nominal barrel diameter of 3" and zero stacking factor). This upper limit is unnecessarily long and can happily be reduced by a factor of four with the added advantage of making the loading the paper stock (rolling on manually) more manageable. A radial thickness of 1/8" of paper provides an unbroken run of about 24 feet which is a practical length and will serve to pursue the ball-park design calculation.

A 24-foot run of paper corresponds to a nominal 31 turns on the feed drum. The size of the annular channel for the clock spring is seemingly too small to house a spring of sufficient length to allow 31 turns for pre-tensioning. However, closer examination of the operation of the take-up mechanism shows that the total number of turns of the take-up spool to transfer all the paper stock from the feed roll is only a small fraction of the number of turns of paper on the roll at the start. When the feed pawl engages during the latter part of the return stroke the feed drum rotates clockwise against the stationary take-up gear so as to issue a short length of paper. The downward arc of the printing roller wraps a length of paper around the take-up drum.

At the start of the run the diameter of the feed roll will be greater than that of the take-up barrel; when the paper stock distribution is equal the diameters will be equal; and at the end of the run the diameter of the take-up drum will be greater. If the take-up drum was not free to move then the incremental length of paper issued each cycle at the start of the run will be slightly longer than that wrapped around the take-up drum during the return stroke. Similarly, if the take-up drum was not free to move then the incremental length of paper issued at the end of a run will be slightly less than that wrapped around the take-up drum. The take-up drum therefore only takes up the small differences that result from the incrementally changing diameters as the paper stock transfers between drums. At the start of the cycle, the compensatory motion from the clock-spring will be clockwise; when the paper stock is equally distributed between the drums there will be no compensatory motion; at the end of the paper supply, the compensatory take-up motion will be anticlockwise.

The worst case difference occurs at the start and end of the paper supply when the distribution is most unequal. For the purpose of estimating the number of turns required of the take-up spool assume that this worst case situation persisted throughout the paper transfer. The difference between the length of paper issued and wrapped for a paper stock 1/8" thick in one turn of the feed drum will be II x(difference in diameters) = II/4 = 0.785" per turn. Since the direction of compensation reverses at the half way point of the stock transfer, the number of take-up turns required is half the total number of turns i.e. 31/2, say 15 turns. The worst case length discrepancy over the whole 24 foot roll will be 15x0.785"= 12" (say). The number of compensatory turns by the take-up spool in either direction will therefore be 12"/( II x3")=1.27 turns. This is well within the capacity of the clock-spring. This estimate assumes worst case diameter difference of 1/4" for the whole transfer. The reality is much more favourable than this. The upshot of this analysis is that the clock-spring is not the limiting factor in the take-tip capacity of the mechanism. Due to the wrapping action of the downstroke the relationship between feed turns and take-up turns is not one-to-one as first appeared.

For a 1/8" thickness of paper stock the line feed with a fresh roll is II D/54 = 0.189" where 54 is the number of ratchet teeth. The worst-case difference in line-height in the printed result (the difference in line height between the start and end of the 24 foot roll) due to the shrinking diameter of the feed drum as the paper stock depletes is (2 II /54)x(thickness of paper stock) = 0.0145". Given that the ratchet advances one tooth per cycle this corresponds to about 8% of the nominal line feed.


Locks are provided at five separate points in the transmission train: the figure racks (176, 174), the lower sections of the vertical racks (172, 173), the printing sectors (172, 173), and both sets of stereotyping sectors (172, 173). In all instances locking is by means of a single common locking bar engaging with an aligned set of v-shaped notches. The locking bars for the sector racks, vertical racks and printing sectors are operated by pivoted levers driven directly by cam followers.

The horizontal racks driven by the figure wheel sectors are shown in plan in 176 (bottom left) and in elevation in 174 (top right). These views show a single wedge-shaped lock serving the vertical stack of thirty racks. The lock is lifted and withdrawn by the angle crank turning on fixed pivot 1V. The crank is driven by link 6H (174) which is in turn driven by the cam follower rocker arm from conjugate cams 2A2, 2A3. The axis of the pivots at each end of link 6H are at right angles to each other. This practice is questionable but the assumption is that the deviation from linear drive is sufficiently small to be accommodated by play in the pivot. 174 does not show the upper section of the lock and therefore details of how the lock is constrained to the vertical during operation are not given. To guide the lock the twin-armed lock lever (4V1, 4V2) and mounting bracket shown in plan in 176 were duplicated in the upper section but the drive link, 6H, was not extended. The motion of the lock is therefore fixed by the radial locus of the upper and lower lever arms only the lower one of which is driven.

The locking mechanism for the stereotyping sectors requires additional explanation. 172 shows what appears to be a claw-lock mechanism for the small stereotype sectors (172 bottom left). However, despite appearances only one locking bar is active during locking. The bars close simultaneously and depending on the position of the type wheel, one or the other of the bars will engage with the v-notches between the type heads. In the position shown in 172, the left hand locking bar does not engage with the sector gear teeth. The lock is operated by raising the twin-tyned locking slider. The angled shoulders of the slider drive the two upper curved followers outwards and bars close in a rocking action. Lowering the slider operates the lower two curved followers to release the lock. The locking slider rides on a single dovetail slide (1Y2, 173) and is driven by forked lever 3X pinned to live rocker shaft 3G The rocker shaft is driven by conjugate cam pair (2A11, (2A12 (173) and contact followers 3W1 and 3W2 (172, 173).

The claw-lock is not redrawn for the large stereotype sectors though it is clear from the provision of the dovetail slider, cams and followers that the method of locking the large stereotype sectors is repeat of their smaller counterparts. The cams in this case are (2A13 and 2A14 and the followers, ?W.


173 shows the cam stack in plan and partial elevations and cam profiles are given in 172 and 175. The cam shaft drive wheel (2A7) is driven from the main drive shaft via an idler (163). The cam stack consists of fourteen cams some of which are conjugate pairs. (This excludes the additional eccentric to provide drive for the inking rollers.)

Figure Rack Lock

The wedge-shaped lock is lifted and withdrawn by the angle crank turning on fixed pivot 1V. The crank is driven by link 6H. (174) which is in turn driven by the cam follower rocker arm from conjugate cams 2A2, 2A3. The cam profiles and follower arms are shown in 172 as part of the general arrangement and abstracted more clearly in 175.

Paper Drum Lift

The linkage driving the printing assembly is shown in elevation in 172 and is described above. The heart-shaped cam (2A4) omitted in 172, and fixed-angle follower (H) is shown clearly in 175 right The v-shaped roller follower pivots on rocker shaft 4B. The cam pushes the roller out to the right during the lifting stroke. The return stroke is not actively driven but the printing mechanism is prevented from being stranded past top dead centre by the action of the spherical counter-weight, 7C2 (bottom right 172), which drives the return stroke which is controlled by the pressure of the follower on the retreating profile of the cam.

Printing Sector Lock.

The printing sector lock cam (2A5) is the T-bar cam shown at about 3-o'clock on 175 right with the follower arm and wedge-shaped locking bar top right of the same view. The follower and locking bar are duplicated in 172 but without the T-bar cam. The T-bar strikes the follower arm 8P and drives the live rocker shaft 8L (173) to which it is pinned. The locking bar assembly 8Q also keyed to the rocker shaft, is driven so as to drive the locking bar into the run of notches to immobilise the printing sectors. The drive is unidirectional and the assumption is that the lock is disengaged by the counterweight of the mechanism to the left of the rocker shaft after the T-bar has passed. Since the T-bar is only a partial cam the rest position of the follower arm is determined by which side of the pivoted mechanism is heavier. If the follower-arm assembly is heaver then the rest position will be the curved section of the follower arm resting on the rocker shaft for the stereotype sectors follower. The likelihood, however is that the locking bar side is heavier and the lock will bounce on the sectors as they rotate.

Vertical Rack Locks

A T-bar cam (2A6) is shown at 6-o'clock in 175 left. The follower arm and locking bar are shown in elevation in 172, 175, and in plan on 173. (The bar is obscured in the plan view and is shown as a dotted outline.) The T-bar acts on the contact follower to engage the lock. The lock disengages by dropping, out or being kicked out when the rack next moves.

Stereotype Sectors

The small stereotype sectors are raised and lowered by conjugate cam pair 2A7, 2A8 (173) via cam followers 8D1, 8D2, shown in elevation in 172. Bosses for the two forked end levers (8D3, 8E) are pinned to the live rocker shaft. The forked end levers raise and lower the stereotype assembly which runs in dovetail slides 5U, 5V (173).

The arrangement is duplicated for raising and lowering the large stereotype sectors. The conjugate cam pair in this case is 2A9 and 2A10. Profiles and followers are shown in 175 and 172 and the forked end levers drive the assembly in dovetail guides as for the small stereotype sectors.

Stereotype Sector Locks (see also page 68 for general comments)

The twin-tyned locking slider (172) rides on a single dovetail slide (1Y2, 173) and is driven by forked lever 3X pinned to live rocker shaft 3G The rocker shaft is driven by conjugate cam pair 2A11, 2A12 (173) and contact followers 3W1 and 3W2 (172, 173). The claw-lock is not redrawn for the large stereotype sectors though it is clear from the provision of the dovetail slider, cams and followers that the method of locking the large stereotype sectors is repeat of their smaller counterparts. The cams in this case are 2A13 and 2A14 and the followers, ?W.

Ink Roller Cam

The ink roller cam (2A16) follower and drive links are shown in plan in 173 bottom centre and in elevation in 172. The cam arm sweeps against the scythe-shaped contact follower which rocks on the fixed pivot working against the action of the compression spring, 1F. A long link operates rocker shaft 1K which operates the two ink roller arms 1K2 and 1K1. The travel of the inking roller towards the print heads is gradual and the restoration to home position is a snap action as the follower leaves the cam and the compression spring is released. The spring has two functions: it provides contact pressure between the inking roller and the sliding roller while in the home position; it restores the inking roller after the downward sweep to the type heads.



Comparison with printing machinery of the time indicates that at least one roller in the inking train was imparted with reciprocal axial motion to ensure uniform spread of ink. The combined rotational and sliding motion was intended to eliminate dead spots which would remain as uninked bands on the circumference of the rollers. A further standard practice was to alternate the consistency of the rollers: the finaking roller (I) would be soft to ensure uniform inking of the print heads, the ink pot roller (Z) would be hard so as to aid sealing against the scraper by providing a true surface, and intermediate rollers would alternate in consistency hard-soft-hard etc. Hard rollers were typically made of steel, and soft rollers from a toffee-like composition.

The method adopted for imparting sliding motion to the roller was to fix two cams at each end of the roller shaft (Assembly Drawing K25). The cams are circular 360o cams with sections of left- and right-handed helices (detail K42) machined as one piece with the roller. The cams ride against two pegs fixed to a support bar above the roller. The angular position of the cams and position of pegs is such that the sliding roller is alternately driven left by one cam and right by the other as a consequence of the roller turning. The roller is driven by a separately by gears i.e. does not rely on rolling friction to rotate.

The possibility of eliminating the two intermediate rollers (K, J) and imparting sliding motion to the final inking roller (I) directly, was excluded: the plan view (173) makes no provision for sliding motion of the inking roller; there is in any event no room for the helical cams between the swinging levers supporting the inking roller. There is therefore the need for a separate sliding roller. This in turn entails the need for an additional transfer roller to maintain the practice of alternating soft and hard rollers. This reasoning provides a credible interpretation of the original roller arrangement shown in 172: the two intermediate rollers were therefore retained in the interests of both authenticity and practicality.

Since the final inking roller is soft, roller K should be hard and I [J?] soft. Since the soft roller has a less determined surface, is subject to greater wear, and in addition needs some pressure to ensure even surface contact on the two rollers it couples, the soft roller should have a floating bearing. Since the helical cam method of imparting sliding motion relies on a fixed relationship between the roller and the axis of rotation, K was selected as the sliding roller (steel), and J as the transfer roller (composition).

The assembly of the inking system is shown on K25. The cluster of rollers is supported by two cast mountings added to the castings for the printer framing members (detail K416 A&B). The detail of the mountings is copied from the mountings shown for the paper feed take up roller immediately below. The transfer roller (J) rests in a cradle which is part of a rocker mounting (detail K436). The roller has a counter-weight (K439) which allows the roller to bear with constant pressure on rollers K and Z. Raising the counter-weight lowers the transfer roller clear of the other rollers and the cradle arrangement allows the roller to be lifted out for cleaning. The ink blade or scraper (K419) is mounted on bronze side cheeks (K418 A&B) which provide the side-walls of the ink reservoir and also seal against the curve of the roller. The side cheeks insert into slots in the mounting and the ink feed is adjusted by adjusting two jacking screws (K423).


A review of contemporary inking arrangements indicated that inking would be too heavy if the ink feed was continuous. Intermittent drive for the ink pot roller was derived from an eccentric added to the main printer cam shaft (D, 173) outside the inking roller drive link (bottom of 173). The arrangement is shown in assembly drawing K25 top left). The reciprocating motion of the link operates a pawl which advances a ratchet wheel (K447) mounted on the roller assembly mounting plate K416A. The amount of rotation of the ink pot roller during each printing cycle is adjustable by altering the position of the sliding pivot on the slotted lever (K446) using the knurled thumbscrew. This alters the pull-back of the pawl and the motion transmitted to the roller. The ink pot and sliding roller are actively driven from the ratchet wheel via a short shaft and gear wheels. Reverse rotation is prevented by a backstop pawl, K448. The other rollers are slaved i.e. do not have independent drives.

Pinion Offset

The plan view 173 clearly shows alternate racks and pinions in the upper section of the rack assembly to be offset. The supposed purpose of recessing alternate pinions between a pair of racks is to trap alternate pinions in the guide channels formed by the racks on either side so as to prevent them fouling adjacent racks in the event of drift in the spindles. The length of the pinion bushes, shown pinned to the spindles in two positions in 173, are about 2.5". Only the first two of these are shown and these are outside the stack assembly. With the arrangement as drawn there is barely any clearance between the bushes of the recessed pinions and the adjacent pair of rack that form the cheeks of the guide. This is further illustrated in 172 centre to which shows virtually no clearance between the cross-section of the spindle and the teeth of the adjacent rack. To provide adequate clearance would reduce the wall of the bush to 1/16". This was regarded as inadequate.

The solution adopted was to abandon offsetting alternate pinions. The double thickness of the teeth (1/4"), created by lapping, allows two measures sufficient to avoid risk of fouling: chamfering the pinion teeth, and slightly reducing the pinion width to provide clearance from the sidewalls formed by the racks. Eliminating the pinion offset removes the need to scallop out the vertical framing piece to accommodate the first few of the lower most pinion bushes until the stagger takes the bush clear of the framing member. The need for scalloping the frame member is as shown for the first pinion in 172.

The lower rack sections which mesh with the printing sectors do not have the benefit of lapping and the rack teeth only l/8" across. Here the offset feature was retained to provide sidewall guidance for the printing sectors. The benefit of offsetting is to reduce criticality of manufacture and fitting by providing guide channels. In the case of the pinions, sufficient margins of clearance are available to justify dispensing with this feature to resolve the problem of wall thickness of the pinion bushes: in the case of the printing sectors, the offset feature was retained.

Printing - Design

Fig. 7 in 165 labeled 'Elevation' is the view of the printing assembly in the fully raised position, shown solid in the companion 'End View' (Fig. 6) alongside i.e. when the paper support roller is pressed onto the line of type. In Fig. 7 the take-up barrel is shown below, the feed barrel and paper support roller above and the feed pawl shown separately bottom left. The take-up barrel is the more complex assembly and consists of a series of sleeves and drums rotating around each other. Two sleeves, integral with the left- and right-hand swinging arms, are pinned to the shaft (7D). The sleeves pass over the shaft and are shown meeting in the middle. The outer surface of the two sleeve halves provide a bearing surface for the next sleeve which has the ratchet and cam fixed at the left hand end and the drive gear at the right Bearing on two shoulders of this ratchet sleeve is a thin-walled outer drum. The drive between the ratchet sleeve and the outer drum is via a clock spring housed in the annular chamber bounded by an end-plate disc and the chamber wall, 4H. The clock spring provides a take-up tension for the outer drum.

Several details are omitted. No method of fixing the cam, ratchet or gear to the single sleeve is shown and therefore the intended order of assembly of the nested sleeves, and the sequence of operations by which the ends of the clock spring are to be secured, if originally considered, are not known. No provision is made for loading the paper stock onto the feed barrel. The feed barrel is not detachable (172, 173) so loading using prepared stock on hollow formers was clearly not intended. The paper would need to be wound onto the feed drum from an external magazine. However, no means of turning the feed barrel is shown, and no method of fixing the ends of the paper roll to the feed barrel or to the take-up spool is indicated. The feed barrel would be wound anticlockwise during loading. Since the feed barrel is geared to the take-up spool, the feed pawl and back-stop pawl would need to be released during loading and there is no provision made for this.

Mechanisms for releasing the backstop pawl, loading paper stock from a magazine, and securing the two ends, were therefore added to the original design. The general assembly is shown in K24 and the with details given in K481A.

The preferred method of securing the paper is by a cross-clamp set as a tenon into an open slot running across the barrel. The paper is trapped under the clamp which is fixed at each end by two screws let into cast lugs in the barrel. The curvature of the outer surface of the clamp follows the barrel radius so that the cylinder of the spool, interrupted by the mortise slot is restored with the clamp installed. The mounting lugs are mirrored on the opposite side of the barrel for balance.

Accommodating the cross-clamp entailed minor changes in some axial and radial dimensions of the assembly as well as modification to the assembly of the clock-spring chamber. Convenient machining the mortise for the cross-clamp requires runout off the end of the barrel. With the clock spring chamber integral with the barrel, the cross-clamp would form a segment of the chamber outer wall. This would expose the spring when paper stock is replenished and run the eventual risk of fouling the spring through the accumulation of paper dust and other debris. Instead of keeping the chamber and barrel as one as shown in 165, it was thought preferable to assemble the clock spring, chamber and gear as a separate unit Separating the spring chamber simplifies manufacture of the main chamber of the barrel, and has the advantage of allowing runout when machining the mortise slot. Stopping the clamp short of the spring chamber prevents exposure of the spring that would otherwise occur when the clamp is removed for paper loading. The arrangement has the additional advantage of allowing the spring, gear and spring housing to be preassembled; the spring is rivetted at one end to the outer wall, and slotted at the other end into the inner sleeve. A further incidental benefit of separating the spring chamber is that with the cross clamp removed, the side of the spring housing acts as a location reference for positioning the paper during clamping.

The clock spring chamber is driven by a pin let into the barrel and protruding into the spring chamber wall. The larger diameter of the pin is secured in the body of the barrel with by an interference fit; the reduced diameter is a clearance fit into the chamber wall. The shoulder formed by the reduced diameter prevents the pin drifting into the spring housing.

With a type pitch of 1/8", 30 printed digits and 1/4" margin each side, the width of paper roll is 41/4". To allow clamping across the full 41/4" paper width and at the same time retain the original spacing of the two swinging arms, the clock spring chamber was crowded closer to the gear than as shown in 165 so as to clear the right hand edge of the clamp. The width of the clock spring and the size of the chamber remains unchanged.

The assembly of the ratchet sleeve is not clear from 165. The ratchet is shown as one with the sleeve; the gear is shown separate but no means of fixing to the sleeve is shown. The ratchet teeth need to be hardened. For ease and economy of manufacture the ratchet is made as a separate item from high grade hardenable steel and secured to the standard steel barrel with three screws and the gear is keyed to the ratchet sleeve. The ratchet wheel is therefore replaceable as a separate item. At the ratchet end, the steel sleeve runs on the cast iron boss of the swinging arm which is an approved match of materials.

To allow room within the original diameter of the outer barrel for the key and for the fixing lugs which secure the clamp, the arrangement of nested bearings was modified. The original design shows the outer barrel bearing on the ratchet sleeve and the full length of the ratchet sleeve bearing on the swinging arm sleeves which are pinned to the shaft and meet in the middle. The altered arrangement is shown in K24. Since there is no absolute need for the

swinging arm sleeves to sheath the full length of the shaft, these two half-sleeves were dispensed with and reduced to a boss at each end, integral with the side arms, and pinned to the shaft. In the original arrangement the inside of the ratchet sleeve bears on the swinging arm sleeve along its full length. In the modified arrangement the swinging arm sleeves are shortened to bosses and the ratchet sleeve is supported by bearing surfaces at each end only: the boss is extended through the swinging arm at the ratchet end to provide one bearing surface; at the gear end the boss does not extend into the barrel chamber the second bearing is provided by a phosphor bronze bush inserted as a press fit to avoid a steel-to-steel bearing surface. Eliminating the two full half-sleeves allows the diameter of the bearing boss at the ratchet end to be reduced from 1-3/4" to 1-3/8". The extra internal space gained by eliminating the sleeves allows the cross clamp securing lugs and a keyway for the gear to be accommodated within the original outer dimensions of the barrel assembly.

The take-up barrel and the clock spring chamber are assembled as a sandwich between the two swinging arms which act as side pieces. The gear forms the outer radial wall of the spring chamber instead of the end plate (165) which was dispensed with.

The cross-clamp arrangement for securing the paper to the take-up spool is duplicated on the feed spool but without the complication of the clock-spring chamber. In this case the face of the gear boss serves as a location reference when loading the paper roll.

Loading Paper

In a relatively rare explanatory annotation we find the following inscription on the drawing for the feed drum: "printing paper coiled on this barrel at the commencement" (165, Fig. 7). Babbage is not forthcoming about how this is to be accomplished and special provision for loading was made. The arrangements for loading the feed spool from a magazine are shown in general assembly K24. Paper stock is supplied from a detachable magazine (K58). This consists of two side cheeks, separated by three spacers, a loose top roller and two wooden cradles which support the paper roll. It is not clear from 173 or 165 whether the printing roller is intended to rotate or not. The preference was to provide a fixed bar with a flat so as to act as an anvil for the type heads. The magazine is lowered onto the printing mechanism and locates in two additional slots machined into the paper support bar which is not free to turn.

The feed spool is wound anticlockwise during loading from the magazine. Since the take-up spool is geared to the feed spool, both the backstop pawl and the feed pawl need to be disengaged to allow free rotation of the barrels. The feed pawl is disabled by halting the calculating cycle before the printing mechanism is fully lowered i.e. before the feed pawl cam releases the pawl for engagement No additional mechanism is required for this. There is no such advantage to be had when it comes to the backstop pawl which remains engaged throughout the normal calculating cycle. Here an additional release mechanism, the pawl release lever, was provided.

The pawl release lever (detail K591 ) is loosely mounted on the ratchet sleeve with clearance between the swinging arm and the ratchet gear. Existing space was sufficient to accommodate the lever without modification to the layout The operating ramp lifts the pawl out of engagement when the lever is operated. The motion is constrained by a pin in the swinging arm which enters a travelling slot in the lever. During loading the direction of rotation (clockwise) biasses the lever to hold the pawl in the disengaged position . Paper is wound onto the feed roller from the magazine by turning the balanced handle rotates the roller. The paper is cut by drawing a blade across a vee-notch cut into the upper spacer on the magazine.

The paper loading sequence is as follows:

  1. Halt cycle with printing mechanism lowered as far as possible without engaging the feed pawl.

  2. Attach loaded magazine to printing mechanism by dropping into slots.

  3. Thread paper past upper spacer, over loose roller on magazine, and under paper support bar.

  4. Clamp end to feed roller using cross clamp.

  5. Lift backstop pawl by turning release lever clockwise and wind paper onto feed roller.

  6. Release backstop pawl by returning release lever to home position.

  7. Cut paper along length of upper spacer and feed around paper support bar.

  8. Tension the take-up spool by rotating one turn clockwise and secure end using cross clamp.

  9. Remove magazine.

Counterbalancing Vertical Racks

Each of the vertical racks has the run of teeth for the mating pinion displaced by the vertical pitch of the figure wheels along a steep diagonal (174). Though the effective section of gearing does not extend the full length of the racks all thirty vertical racks are assumed to be the same length and run the full height of the engine. (The only view of this is on 174 which shows the lower section only.) The racks are made of steel and their collective weight is 41.4 lbs. The racks, which are in loose contact over their length and will be lightly coupled through sliding friction. The effect of this is increased by the L-shaped lapping of adjacent pairs of racks (173).

There is no evidence in the original design for counterbalancing the deadweight of the racks and there are two points in the cycle where there is a risk of runaway i.e. of uncontrolled downward motion as the racks lower themselves under their own weight The first possible runaway condition occurs during the transfer from the last even sector column to the figure racks. The transfer from odd to even columns in the first half-cycle sets the tabular value on the eighth column. The tabular value is transferred from last even sector column to the figure racks (176) by the last even sectors during the second half-cycle i.e. during even to odd addition. The second half-cycle commences with the tabular figure wheels in general displaced anticlockwise from zero, even sectors fully raised (disengaged) and figure wheels locked. The even sectors then fully lower to engage, figure wheel locks withdraw, and figure wheel axes and zero stops lower. The figure racks are unlocked throughout this sequence. The expected action here is that the figure wheel axis internal arms will drive the figure wheels to zero, and in so doing drive the figure racks via the sector wheels in a controlled manner. However, when the figure wheel locks withdraw there is a danger that the racks lower under their own weight and run ahead of the figure wheel drive arms. The effect is for the vertical racks to drive the train (figure racks, sectors and figure wheels) until the outer nibs of the figure wheels ram against the zero stops. The worst-case condition occurs with all figure wheels set at '9'. Here all racks fall the full vertical travel of 2.827 inches. The trains for figure wheels at positions intermediate between '0' and '9' will be driven in proportion to the figure wheel value; figure wheels set at zero are unaffected.

The coupling of racks through light sliding contact will have two effects: one effect is for less mobile racks to act as a brake for runaway neighbours; the second effect is to avalanche process in which the less mobile racks join the downward motion through lateral coupling.

The runaway does not introduce either calculation or printing errors. The effect is simply to set the figure racks up with the desired tabular value in an uncontrolled rather than controlled manner, and to impact the zero stops.

The second risk of runaway occurs during the restoration after stereotyping. Once the tabular value is transferred to the printing/stereotyping heads the figure rack locks and figure wheel locks secure the train and immobilise the mechanism during printing and stereotyping (385a). The locks then disengage and the intended next action is for the sweeping sector arm to engage with the pegs on the upper side of the sectors to restore the tabular value, return the even sectors to zero, and restore the figure racks. It is at this point that the second runaway condition is liable to occur. The general condition is for the sectors and figure racks to be displaced by different amounts from zero. With the locks removed and the figure wheel zero stops raised, the racks will tend to lower again under their own weight and drive the train in reverse. The limit of the motion this time is not the figure wheel zero stops but the advancing sector wheel restoring arm which will impact the drive pegs.

Neither of the two runaway conditions introduce calculating or printing errors. In the both cases the restoring action of the drive arms will complete correctly. However, the ramming of the mechanism against the zero stops in the first case, and against the rotating sector restoring arm in the second, is regarded as undesirable. There is also concern about the load of the racks on the sector restoring arms when the racks are lifted after printing as well as the overall load on the manual drive when the racks are collectively lifted and restored to zero.

To avoid the risk of runaway the racks are counterbalanced by providing individual counterweights for each rack. 174 shows 30 racks 1/8" wide and spindles nominally 1/4" wide. It is evident from the plan view (173) the arrangement is more subtle: the L-shaped lapping discussed earlier means that the 15 odd-numbered racks face the front (digits 1,3,5, ... , 29) (the least significant digit represented on the right-most rack), and the 15 even-numbered racks face the rear. The L-shapes provide a vertical face 1/4" wide for the rack teeth and for meshing with the 1/4" pinions. 174 has solid lines at 1/8" pitch to represent the racking members and the section with the racking teeth 1/4" wide. This indicates that the only sections that are lapped are the toothed section of each racking member; if the lapping was extended beyond the toothed section the alternate vertical lines would be broken. However, the use of broken lines to show hidden edges is not consistent elsewhere, and drawings are sometimes an inconsistent mix of sectioned and diagrammatic views. Licence was therefore taken to extend the lapping beyond the toothed sections upward and downward but clearly stopping short at the lower end where the printing sector racks revert to a true 1/8" pitch i.e. with no lapping flange. Details of the racks are shown on K332 A-Q.

Contemporary practice indicated that if the racks were to be counterbalanced this would likely to have been achieved by individual weights suspended by cords wound on drums in the pinion shafts. Two factors weighed against this arrangement: the lowermost pinion shafts have insufficient clearance for the full downward travel of the weights; and the counterbalancing load is transmitted through the friction of the shaft bearing and pinion gearing, and this force is variable. The arrangement preferred involves individual counterweights being attached directly to each rack by a cord suspended over a pulley on a shaft above. The weights are long steel strips (8" x 1 1/2" x 1/2") and weigh 1.38 lbs each. Each of two pulley shafts run left to right (front and rear) carry 15 3-inch diameter phosphor bronze pulleys (K322) 1/4" wide. Viewed from the left hand end of the engine the cords are centre-parted evenly to the front and rear.

Adjacent weights are in light contact and inhibit any tendency to spin. The weights are suspended in two tiers i.e. 15 weights in two tiers at the front (eight above and seven below) and 15 in two tiers at the rear. Channels cut down the faces in contact allow paths for the cords to the lower tiers. The arrangement is shown in K22 and K21. The top of the racks are cut away and drilled with 1/6" holes for the cords. The purpose of the cut away is to ensure central suspension points to avoid listing off the vertical.

There was concern that the cords would slacken momentarily when the racks were raised by the pinions, with the danger that the cords would jump the pulley wheels. A brief calculation reassures otherwise. For a ten-digit restoration of the rack the rack is raised 2.827" in 60o of the cycle. In the first 0.01 sec the free-fall distance of the weight is 0.019" and the rack movement is 0.028" i.e. 0.009" cord slack. The pulley channel depth is approximately 0.0625" i.e. the initial slack is insufficient for the cord to jump the pulley.

An additional benefit of counterbalancing is the reduction of wear on the figure racks locks and figure wheel locks. As the engine wears, the corrective action of the locks increases. Without counterbalancing, the locks would be inserted against the weight of the racks. This additional load would accelerate wear of the locks as well as increase the shock-load on the cam. This is regarded as undesirable given the steep pressure angle of the lock cam profile and the inability to turn the engine without counterbalancing of the locks with additional springs. The provision of counterbalances for the racks was considered to be consistent with the practice of providing spring counterbalances in the original design for the figure wheel shafts (163).

Go to next section Stereotype Table
Go to top of Babbage Technical Description