+Home     Museum     Wanted     Specs     Previous     Next  

Singer/Friden 1152 Desktop Calculator

Updated 2/9/2020

One weakness that most early electronic calculators had was that the vast majority of them utilized some type of visual display device, examples being Nixie tubes, gas-discharge tubes and panels, CRT displays, vacuum-fluorescent tubes, and incandescent indicators. The problem with all of these indication technologies was the lack of hard copy output. For this reason, some realms of calculating that could benefit from the speed of electronic calculators actually kept on using their "old" electro-mechanical printing calculating machines. Why? While slow and noisy, these mechanical calculators provided a permanent record of their calculations. For financial institutions, accountants, and bookkeepers, a hard-copy that could be saved as a check of the math performed was a very important requirement.

Calculators that use a visual display to indicate their answers had the disadvantage that results had to be manually copied onto some form of permanent record, making the chance for human transcription error much greater than a calculator that provided a printed output. A means had to be developed for an electronic calculator to be married to some type of printing technology to provide the best of both worlds...electronic speed, and hard copy output. One of the biggest challenges was building a printing mechanism that had sufficient speed to minimize the impact on the speed of the electronics, and wasn't so expensive that it priced the machines out of the market.

While Friden had enormous success with their first-generation display-type electronic calculators, the Friden EC-130 and Friden EC-132, it was clear that a segment of the market wasn't buying into calculators with displays. There were a few electronic printing calculators on the market at the time, such as the Monroe EPIC 2000 and EPIC 3000, and the benchmark-setting Mathatronics Mathatron, but these calculators were expensive, high-end machines with scientific functions and programmability. Wang Laboratories also offered printing peripherals for their calculators, but, like the Monroe and Mathatronics machines, these machines offered functionality beyond the needs of basic business accounting. Bankers and accountants weren't interested in paying the extra money for these advanced features when they would never be used. Wanderer Werke in West Germany had developed a fairly advanced programmable electronic printing calculator, but their market was mainly in Europe. Friden needed to address this market segment with a four function electronic calculator that made a printed record of its calculations, and was cost-competitive with display-type calculators.

Friden 1152 Internals

The Friden 130 and 132 were becoming a bit dated by 1966. The machines were large, heavy, relatively slow, and expensive compared to some of the newer calculators that were showing up in the rapidly expanding marketplace. The Japanese had quickly developed the technology to make electronic calculators that had similar capabilities to American and European-made calculators at markedly lower retail prices, even after export to these markets. Robert Ragen, the brilliant engineer that was the mastermind of the Friden 130 and 132 calculators, had designs in mind that would utilize the same basic architecture of the 130/132 transistorized calculators, but would implement the logic using the new technology of integrated circuits. An IC-based design would save space and weight, along with reducing power consumption. Such savings make the machines less expensive to manufacture due to reduced parts count and complexity. Ragen set about developing designs for two lines of calculators that would share the same general design, differing only in their means of communicating results to the user. One line of machines would utilize a CRT-display, similar to the 130/132 calculators, and the other line would consist of printing calculators. The priority would be put on developing the printing calculators first, specifically to address the market need for a Friden printing electronic calculator, and the IC-based display calculators, though mostly developed in tandem with the printing machines, would be introduced slightly later.

A new feature would also be implemented in some models within each line -- simple programmability, a feature likely included due to it being included in calculators from domestic competitors as mentioned earlier. Users who frequently ran repetitive calculations could use this feature to automate the calculations, making only variable entry required by the user, with the calculator carrying out the learned sequence of mathematical operations automatically. To Ragen, the addition of printing capability, along with the added utility of programmability, seemed to be just what the market needed to address the shortcomings of the earlier calculators.

Model Identification on Back Panel (note massive heatsink)

While the IC-based design of the calculators reduced the component count and circuit complexity versus their transistorized predecessors, the printing calculators ended up adding cost and complexity through the use of a complicated printing mechanism that somewhat offset the benefits of the use of integrated circuits. The result was that the printing calculators were still rather expensive, as well as being rather large and heavy. On top of that, the stack architecture that the calculators used, while certainly superior in terms of its ease of use for complex calculations, was not something that accountants and bookkeepers were used to...and these people were the folks who were most likely to buy a printing electronic calculator. The bottom line was that while Friden's first electronic printing calculators were of Friden's typical quality and reliability, they somewhat missed the mark in terms of addressing the primary requirements of the intended market, business and financial users, whose main concerns were of course, printed output, but also low-cost and ease-of-use.

Given the fact that the Japanese had begun creating a strong marketing presence in the US and Europe for their low-cost electronic calculators, Singer's management began to realize that perhaps the design and manufacture of electronic calculators domestically was soon to become a money-losing proposition. While Ragen and his team of engineers were working on developing Friden's second generation of electronic calculators, Singer's management had begun looking into alternatives to address a market that demanded lower-cost calculators. The outcome was that Singer's management forged a partnership with Japanese electronics giant (and electronic calculator manufacturer) Hitachi. The forging of this relationship spelled the beginning of the end of Friden's in-house electronic calculator development.

As an experiment to test this relationship with Hitachi, Singer imported Hitachi's KK-12 electronic calculator, putting the Singer/Friden badge on the machine, and designating it as the Friden 1112. The 1112 was actually marketed by Singer/Friden before Ragen's second-generation calculator designs came to market. The development of Ragen's new series of machines continued while the 1112 was being marketed, but it soon became clear that the broad market was willing to forego printing and programming functions just to have an electronic calculator that was comparatively inexpensive. The 1112 was a reasonable market success, and more importantly, had a decent profit margin, proving to Singer's management that their hunch was right. However, by the time all of this came to light, Ragen's team had a printing electronic calculator design that was nearing production reality. Rather than cancel the project and lose the investment in the development of the calculator, it was decided to go ahead and bring the machine to market to try to recapture some of the development costs. The resulting calculators became the Friden 115x-series of printing electronic calculators, and the Friden 116x-series (See the exhibit on the Friden 1162 for more information on the 116x-series calculators) of CRT-display calculators -- the last calculators fully designed and manufactured by what was left of Friden Calculating Machine Co.

Model/Serial Number Tag

These two lines of calculators all share a similar architecture to the Friden 130/132 machines, utilizing the same four-level RPN stack arrangement, pulse-train digit representation, counter-based arithmetic unit, fixed decimal, and magnetostrictive delay line for working register storage. Primarily DTL (Diode-Transistor Logic) and some TTL (Transistor-Transistor Logic) small-scale integrated circuits are utilized for the logic of the machines, significantly reducing the size and weight of the calculators compared to the 130/132, and allowed additional functionality. The Friden 115x-series calculators were Friden's answer to the perceived need for a printing calculator. The Singer/Friden 1152 calculator exhibited here is the high-end machine in the original 115x-series, providing four functions, thirteen digit capacity, automatic square root, a single store/recall memory register, and 30-step programmability. Three other calculators made up the initial 115x-series. The 1150, the first calculator introduced in the 115x-series, provided the basic four math functions and single memory register, with no programming capability. The 1151 was a 30-step programmable version of the 1150. The 1152, as exhibited here, adds one-key square root to the features of the 1151. Lastly, the 1154 is essentially a re-make of the 1150, using a new-and-improved printing mechanism that is both faster, easier to build, and more reliable, but based on the same design as the 1150/1151/1152's printer. The 1154 eliminated the four LED indicators on the front panel for displaying the status of each of the four registers of the stack. Along with the improved print mechanism, the 1154 was also optimized in numerous ways to reduce its manufacturing cost

While the 115x-series printing calculators did sell to some degree, primarily to devoted Friden calculator customers, as well as also on Friden's reputation for quality and exceptional services, the calculators didn't really sell into the intended marketplace, with many of the machines selling into engineering, scientific, and technical application spaces, rather than the business/financial market that seemingly had the most opportunity as a new market for electronic calculators.

Another calculator was added to the 115x-series in early 1971, some time after the other machines had been on the market. This calculator, while sharing the 115x-series nomenclature and cabinetry, was, a complete architectural and functional departure from the earlier machines in the series. Rumor has it that the the machine was designed and developed by an independent contractor hired by Singer, much to the disdain of Friden's calculator engineering team. The charter of the contractor was to develop a high-end programmable, printing scientific calculator using a modified version of the earlier 115x-series printing mechanism, as well as sharing the cabinetry from the machines to save on development costs. The word is that this machine was the "last chance" that Singer was to allow Friden to have a market success like the 130/132 calculators had. As it turned out, the calculator took too long to develop, coming to the market too late to have any significant impact in the marketplace.

This calculator became the Friden 1155. The machine was an advanced scientific calculator, with much more comprehensive programming capabilities, and significantly more memory registers, not to mention a large repertoire of math functions. The calculator eschewed Friden's pride-and-joy RPN math architecture, instead using pure algebraic logic. The machines utilized large-scale MOS integrated circuit technology, eliminating the magnetostrictive delay line, using MOS memory for storing the working registers, program steps, and memory registers. As specified, the 1155 utilized the cabinetry components and an augmented printing mechanism from the earlier 115x-series calculators. The 1155 provided 20 memory registers, and up to 511 program steps of capacity. Shortly after the 1155 was introduced, an update using higher- capacity storage ICs was introduced, designated the 1155A, which upped the number of memory registers to 100 memory registers, but other than the difference in memory register capacity, the 1155 and 1155A were identical.

The original Friden 1150 printing calculator debuted in February of 1968. The 1152 exhibited here was manufactured in early 1970, based on date codes on the integrated circuits in the machine. The 115x-series calculators were among the earliest American-made calculators to use integrated circuits. The Japanese had been marketing calculators utilizing bipolar and early MOS (Metal Oxide Semiconductor) integrated circuits as early as 1966. For whatever reason, American calculator manufacturers were a bit slow to adopt integrated circuit technology, with some American manufacturers (such as Wang and Hewlett Packard) not introducing integrated circuit-based calculators until the early 1970's.

Like Ragen's previous designs, the 1152 utilizes a four register RPN (Reverse Polish Notation) stack architecture. The four registers are connected in such a way that new entries are placed into the 'bottom' register of the stack. Numbers that go off the 'top' of the stack are lost. The [FIRST NMBR/PRINT] key completes entry of a number, preparing the machine for entry of the next number. As soon as the first digit of the next number is entered, the stack is pushed up one level, and the new number is entered into the bottom of the stack. Addition, subtraction, multiplication, and division operate on the bottom two registers of the stack, popping the stack down one level, leaving the result in the bottom register of the stack. The square root operation calculates the square root of the number in the bottom register of the stack, leaving the result in the argument's place, without disturbing the stack.

The Control Panel (including stack status indicator) of the Friden 1152

Given the lack of a display on the 115x-series calculators, a novel means for indicating the status of the RPN stack was developed. In fact, the concept was unique enough to patent. An indicator on the front panel of the 115x-series calculators shows the status of the stack. The indicator has four incandescent lamps stacked in a vertical configuration, behind a red plastic lens. If a register in the stack contains non-zero content, the corresponding indicator lights. For example, if the bottom and next-to-bottom register contain numbers, the bottom two indicators will be lit. Carl Herendeen, another of Friden's brilliant and prolific engineers, developed the idea for this status display, and was granted US Patent 3,495,221 in February of 1970.

Three stack manipulation functions are provided. The [I] key exchanges the bottom two registers on the stack, and the [DUP] key pushes the stack up one level, duplicating the number of the bottom of the stack. The [FIRST NMBR/PRINT] key actually has three functions. The first function, entering a number into the stack, was documented earlier. If pressed after a math operation, the [FIRST NMBR/PRINT] key will print the content of the bottom register on the stack. If pressed a second time in succession, all four registers on the stack are printed, one register per line, with the "top" stack register printed first, and the "bottom" register last. The [ENTRY/CLEAR STACK] key is another dual-function key, with the first press clearing the bottom register of the stack, useful for erasing erroneous entries. If pressed a second time, the entire stack is cleared. In both cases, the content of the memory register is not affected.

The Keyboard of the Singer/Friden 1152

The memory register of the 1152 is a simple store/recall register. Pressing the [TO MEMORY] key takes the number on the bottom of the stack and copies it to the memory register, popping the stack downward. The [FROM MEMORY] key causes the content of the memory register to be pushed into the bottom register of the stack.

The programming features of the 1152 are rather unique. The machine only stores function key presses. Numeric entries are not stored as steps in the program memory. The concept of programmability for these machines is intended only for simple linear operations. These machines were not designed for solving complex problems through their programming capability. The inability to store constants as part of a program seriously limits the complexity of problems the machines can address. The programming feature of the 1151 and 1152 was intended for automating simple, repetitive types of calculations -- for example, performing discount or mark-up calculations in a retail operation, or calculating the area of circles for an engineering application.

The programming functions are controlled by three keys. The [LEARN] key, which locks down when pressed, puts the calculator into learn mode. In this mode, the calculator operates as normal, but each operation key ([FIRST NMBR/PRINT], [+], [-], [X=], [÷=], [I], [DUP], [TO MEMORY], [FROM MEMORY], and [AUTO]) is stored upon entry into the delay line, and then executed as usual. The steps are stored in the delay line as single-digit numbers, much like normal numeric entry, but by virtue of their location in the delay line, they are interpreted as instructions when played back. The [AUTO] key has a special meaning when recorded in LEARN mode; when this code is encountered while a program is being run, the calculator stops and waits for user input, allowing the user to enter variable data. The [PROG RESET] key clears learn mode (releasing the [LEARN] key), and begins execution of the just-entered sequences of operations. The [AUTO] key resumes execution when the machine stops awaiting variable data from the user. It should be noted that while the 1152 has a square root function, it is not possible to use the square root operation within a program. If the square root key is pressed in LEARN mode, the function will be performed, but it will not be stored in the program memory. This is because there are only ten possible operation codes that can be stored in the program memory, and all ten codes are used by the basic functions of the machine. For more information on programming the 1152, see the 1151 Programming Instructions manual (programming on the 1152 is identical to that of the 1151.)

Example of Printed Output from Friden 1152 with Annotation

The printing mechanism of the 115x-series calculators is quite unique and ingenious, although rather complicated. It appears that the mechanism wasn't actually designed specifically for printing calculators, but was originally designed as a printer for inventory ticket printing, and was picked up for use in the printing electronic calculators. The design of the mechanism was developed by Friden engineers Leland D. Chamness and Andre F. Marion, who were granted US Patent #3406625 in October of 1968 for the design. Given Friden Calculating Machine Company's famous mechanical and electro-mechanical calculators, it is no surprise that the printing mechanism is a prime example of the mechanical brilliance of Friden's engineers. Interestingly, as late as 1970, the Friden division of Singer Corp. was offering the same basic printer mechanism for sale as a "Digital Printer" module. The unit included all of the drive and amplifier electronics, making it a relatively simple device to interface to. The 11-pound, 11-3/4" wide by 7-5/8" high by 8-3/4" deep device sold in single quantities for $450.

The printer used in the 115x-series calculators is a serial (character at a time) impact printer. Unlike most other printing calculators of the 1970's, which use a drum with multiple columns of embossed digits, and a like number of hammers to print a line at a time (see the exhibit on the Wang 600 for an example of this type of printing technology), the serial means of printing used on the Friden 115x-series calculators is quite unique. Interestingly enough, this design lives on, as many of today's small low-cost desktop printing electronic calculators use a printing mechanism similar to this design.

Printer Mechanism Diagram (from US Patent #3623009)

The printing mechanism consists of a carriage containing a print wheel that is embossed with symbols around its periphery. The print wheel can move horizontally on a shaft that rotates, turning the print wheel at high speed. The symbols embossed on the print wheel are arranged in two groups. The first group consists of the digits zero through nine and the decimal point. The second group contains function indicator characters, ("+", "-", "=", "C", "F", "X", "M", "E" and "÷"). The carriage carrying the print wheel also carries a solenoid-activated hammer, located behind the print wheel. The hammer is situated such that when the solenoid is activated, the hammer moves toward the print wheel to strike it.

The print wheel showing detail of the embossed numeric section

The carriage is mounted on a slide such that the printing mechanism can move back and forth horizontally. A helically grooved shaft is located below the carriage. The grove in the shaft is machined into it in such a way that when the carriage is mechanically coupled to the shaft, the carriage will be pulled horizontally (to the left) as the shaft rotates. A solenoid-activated pin in the carriage engages the groove in the shaft to allow the carriage to be carried to the left from the home position (located at the rightmost end of the paper) when the solenoid is energized. When the solenoid is released, a spring causes the carriage to quickly return to the home position. The paper to be printed upon is situated such that it is interposed between the print wheel and the hammer.

OEM Replacement Ink Roller

A special replaceable inked roller cartridge (Singer part number 811490) provides a supply of ink, which is transferred to another special rubber roller that rides against the print wheel to keep the embossed characters on the periphery of the print wheel consistently coated with a layer of ink. A toothed metal wheel is connected to the left end of the shaft that rotates the print wheel. The teeth on this wheel are positioned such that they correspond to the characters embossed on the print wheel. As the wheel spins in synchronism with the print wheel, the teeth of the wheel trigger small pulses of current in a series of three coils located in close proximity to the toothed wheel. One coil pulses as the number section of the print wheel is nearing position, another pulses when the symbol section of the print wheel is nearing position, and the last coil pulses as each character on the print wheel is positioned so that the hammer can strike it to cause the character to print. The pulses coming out of these coils tell the electronics that drive the printer which character (be it digit or function symbol) is currently aligned with the print hammer. As the carriage travels across the paper right to left, the signals coming from the print wheel position sensors trigger the hammer solenoid to fire at the correct instant to cause the various digits and symbols to be printed on the paper, character at a time. Once an entire line of characters has been printed, the carriage is released and returns to the home position, and a clutch activates to cause the paper to advance one line, readying the printer for the next printing cycle. A single high-torque electric motor, rotating at approximately 3600 RPM, provides the rotational energy that operates the entire print mechanism. The motor is only powered up when printing (or paper feed) is occurring, limiting the noise of a motor running continuously. The 1152 is rather noisy while printing, with the noise of the motor combining with the "clackity-clack" of the print hammer. The printer prints at the surprisingly fast rate of 37 characters per second, resulting in an average rate of 1-1/2 lines per second. This speed allows the calculator to have the benefits of the speed of electronic circuitry, while still providing a hard copy of the calculations. Given that the special keyboard of the 115x-series calculators locks while calculations are in progress (including the time it takes to print a result), it is not possible for a skilled operator to out pace the machine. Digits in front of the decimal point are printed in groups of three for easier reading. For example, the number 123456.789 would be printed as '123 456.789'. A pushbutton on the top of the cabinet activates the paper feed mechanism to allow the paper to be advanced by the user.

The Printer Drive Circuitry

A circuit board mounted on the left end of the printer mechanism contains the circuitry for conditioning the print position sensor pulses, as well as drivers for the various solenoids that operate the printing mechanism.

The Circuit Boards Used in the Friden 1152 (Click on image for larger view and more details about the board)

Most of the integrated circuits in the 1152 are made by Texas Instruments, with a few chips from other vendors (Motorola, Fairchild and National Semiconductor) sprinkled in here and there. The machine is made up of a mix of very early 7400-series TTL IC's (mostly 7474 dual flip-flops), along with Texas Instruments Small-Scale DTL (Diode-Transistor Logic) devices in the SN158xx series. Along with the small-scale logic, the 1152 (as well as other machines in the 115x and 116x lines) use three medium-scale integration TTL devices made by Texas Instruments. The chips consist of SN1286, SN1287, and SN1288 devices. Each device implements one of the counters in the "three counter" arithmetic unit of the machine. The method of using counters (rather than adders) to perform the arithmetic came from the original Friden EC-130 calculator. In the original EC-130 design, four counters were used to count pulses in order to carry out math operations. As time went on, it was realized that three counters could perform the same function, so design changes were made to the EC-130 and EC-132 calculators to eliminate one of the counters, reducing component count (and thus cost). This arithmetic architecture was quite efficient, and was carried over to the 115x/116x-series calculators, with the implementation of each of the counters done in integrated circuit form.

The Printed Circuit Backplane of the Friden 1152

The 1152 uses a total of 166 integrated circuits, along with a significant accompaniment of transistors, diodes, resistors and capacitors. The main logic of the machine is spread across six circuit boards that are plugged in vertically in a hefty aluminum card cage located toward the rear of the chassis. The circuit boards plug into a backplane made from a printed circuit board, which connects to other areas of the calculator (power supply, keyboard, printer, status indicators) via wires soldered to the pins of the edge connector sockets. The circuit boards in the machine are made of high quality fiberglass construction, with gold-plated edge connector fingers. The boards have traces on both sides, with plated-through feed-throughs to provide connections between traces in opposing sides. Each circuit board is approximately 9 1/4" by 5 1/4", and has two groups of edge connector fingers, consisting of 44 pins per group (2 x 22), for a total of 88 pins.

Power supply circuitry of the 1152

The power supply of the 1152 is a basic linear supply, using a transformer to step line voltage down to various AC voltages, which are then rectified, filtered, and regulated to make up the various DC voltages used in the calculator. The power supply is split up in a number of areas within the machine. A main power supply board is mounted alongside the printer assembly that contains some of the voltage regulation circuits, as well as the large electrolytic capacitors that provide filtering. Another small board located near the delay line in the base of the machine provides some of the specialized voltages needed by the delay line. The two transformers that step down line voltage are mounted to the chassis. Lastly, the large power transistors and high-wattage resistors that dissipate significant amounts of heat are secured to the large heatsink that makes up the back panel of the machine.

The Delay Line (in aluminum housing) Assembly in the base of the Friden 1152

Though IC's had replaced the all-transistor electronics of the earlier 130/132 calculators, the 1152 still uses magnetostrictive delay line technology for storing the working registers of the machine. The delay line of the 1152 is contained in an aluminum housing in the bottom of the machine, as in the 130/132. At the time, low-cost integrated circuits did not have sufficient capacity to represent all of the working registers of the calculator. The magnetostrictive delay line was a tried and true technology, and was also an inexpensive way to provide sufficient storage capacity for the calculator. The delay line in the 1152 contains a total of six registers, designated RS, R0, R1, R2, R3, and R4. RS is the memory register, R0 is a special-purpose register used by the printing system and is not accessible to the user. R1 through R4 are the four registers of the RPN stack. Each register can hold the representation of 25 digits, with an additional digit location (for a total of 26) reserved as a marker to tag the beginning of a register in the bit stream. Each digit is represented by a number of pulses. No pulses during a given digit time represent the digit zero. Nine pulses represents the digit nine. Each digit time is divided into 16 slots, each of which can be filled with a pulse (essentially a '1' bit) or no pulse (a '0' bit). As a result, each register consists of 26 X 16 potential bits, or 415 bits. With six registers stored in the delay line that makes a total of almost 2500 bits that are stored in the delay line. You may have noted that each register has 25 possible digits, but the capacity of the machine is only 13 digits. A total of 15 digit positions store a number in the machine, with two additional digit positions containing codes that indicate the location of the decimal point within the number, and another code that indicates the sign of the number. The remaining 10 digit slots of R4, R3, and RS make up the 30-step program store for the programming feature of the calculator. The extra digits of R0, R1, and R2 are not used, but still circulate through the delay line. One wonders if perhaps these extra digits might have been reserved for future use in expanding the program step storage capacity of the calculator, potentially doubling the program step capacity to 60 steps. While technically it seems feasible that this could have been done, but for whatever reason, Friden did not ever add this capability to the 115x-series machines.

The 1152 calculates at the same basic rate as its CRT-based relative, the Friden 1162. All 9's divided by 1 takes just over one second to complete (not including printing time). Square root operations take an amount of time based on the number of digits in the argument, with calculation times ranging from about 0.25 seconds to almost 1.5 seconds. Printing occurs as soon as the calculation completes. During the calculations, the stack status indicators flicker a bit, then settle into the correct configuration once the calculation completes.

The cabinet of the 115x-series calculators consists of a heavy cast aluminum base, with a stamped metal removable grille covering the bottom of the base to allow service technicians access to the adjustments for delay line timing. The upper portions of the cabinet are also made of cast aluminum. The keyboard bezel is made of a plastic casting. The front panel of the machine is a cast-plastic insert which was available in four different colors to fit in with office environment decor. This insert was available in Slate, Aqua, Olive, or Terracotta. The machine exhibited here has the Terracotta-colored insert.

Front Panel Insert in "Olive" Color

Division by zero results in the machine aborting the calculation, printing an "E" in the rightmost-column and clearing the stack. Overflow causes the same behavior. Extracting the square root of a negative number does not flag an error. The calculator returns the answer as if the argument was positive.


Text and images Copyright ©1997-2023, Rick Bensene.

All content on this site is not to be gathered, scraped, replicated, or accesed in any way for any use in populating machine learning or intelligence (Artificial Intelligence, a.k.a. AI) databases, language models, graphs, or other AI-related data structures. Such use is a violation of copyright law. Any such access will be reported to the Oregon Attorney General and prosecuted to the fullest extent the law allows.