The Breakthrough
At the beginning of 1947, Rameev heard a BBC radio broadcast announcing that United States scientists had built an extraordinary electronic computer consisting of eighteen thousand electronic vacuum tubes, all of which were connected by tens of kilometers of cables. When Rameev heard about the ENIAC, he sensed that this was the branch of science and technology which was meant for him. He decided to discuss this with Berg, who was a very approachable person. Berg recommended that Rameev get in touch with Brook, who had been working with computing technology at the Power Engineering Institute. Brook already had a mechanical integrator-analyzer at his lab, a very bulky analog computer which was very difficult to operate. Back then, the air was literally filled with ideas of building digital computers and Brook was happy to receive an enthusiastic assistant. In May 1948 Rameev was admitted to Brook's laboratory as a design engineer worked directly in one of Brook's own offices.
Incredibly, the first result was achieved only three months later, in August 1948: the Avtomaticheskaya tsifrovaya vuichislitelnaya mashina, or "Automatic Electronic Digital Computer," ATsVM project- the precursor to the M-1. Corresponding Member of Academy of Sciences, I.S. Brook and engineer B.I. Rameev, filed its project report. The following is an abbreviated copy of the original document preserved in the Moscow State Polytechnic Museum.
|
The Automatic Digital Computer (Short description) Corresponding Member of the AS USSR I. S. Brook B. I. Rameev, Engineer Moscow, August, 1948 CONTENTS I. Introduction II. General Description of the Computer III. General Description of Individual Elements 1. Device for preparation of program tape and transferring input data from decimal to binary system 2. Main program controller 3. Identifier of the sign, equality and inequality of two numbers 4. Adder 5. Multiplier 6. Divider 7. Storage 8. Interpolator 9. Device for transferring output results from the binary to the decimal system and for printing on paper. IV. Description of several computer relay elements 1. Magnetic relay with two stable states 2. Magnetic trigger 3. Magnetic relay, which operates only when simultaneously receiving several control signals 4. Magnetic relay, which responds when one control signal enters any input channel 5. Decoder V. Appendix. Table of basic parameters of high-performance digital computing machines designed or under development in America. I. Introduction Recently, some information has surfaced in the foreign press about plans to design and build high-performance digital computing machines. The first machine that became operational in America during the war worked on the impulse-counting principle, where impulses were registered on electromechanical counters. This was a general-purpose computer for solving various mathematical tasks by means of counting finite differences.[3] The computer has a relatively low operating speed and restricted storage capacity (only 60 words). According to the available information, this machine has been widely used along with differential analyzers to solve problems related to the so-called Manhattan Project. After the first computer, a second and purely electronic one – the ENIAC – was developed and used primarily for solving the ballistics problems at the Aberdeen Army Proving Grounds. We will not dwell on the design of this machine or its features, which are only known from brief reviews in the current literature. Neither are we familiar with its principal disadvantages or those of its predecessor – the Harvard computer. Currently, several American organizations are busy developing new and improved computers. They are building a new machine at Harvard, two machines for the Bureau of Standards, and several more for universities, institutes, and special Army and Navy research centers. Similar computer projects have begun in England and France. The current literature also discusses a number of problems that might benefit from the use of such machines. These include the calculation of function tables, astronomy problems, processing statistical data, and preparation of bibliographical reference books. But the main purpose of such computers, which are also very costly to construct, would without a doubt be the solution of scientific and technical problems related to defense and the development of modern forms of military technology. For example, the American Bureau of Standards – an organization analogous to the Soviet Bureau of Weights and Measures – has a large department to investigate problems of artillery control. As reported in other publications, the same problem is being studied at several research firms, laboratories and special Army and Naval research institutions. One of these computers is to be used solely for weather forecasting calculations, which is an extremely important task during wartime. Finally, there is one other area that nobody mentions, but which can definitely benefit from the use of a computer or even a series of computers. It is the field of cryptography, which carries special significance for intelligence work. It is not possible to list every potential application of computers. Therefore, discussion here will be limited to the general trends in modern scientific research and construction that deal with new military technologies. The path from the initial idea to the first prototype is incredibly long. Therefore, it is very important to replace expensive and time consuming experiments with computations. Everybody knows how difficult and practically impossible such calculations are, even when the task can be expressed in exact mathematical terms. In addition, the accuracy of the results needs to be extremely high because the absolute error for the size of the values that we are working with (for example, the high speeds and great distances in artillery control), must have a very narrow margin. Such problems cannot be solved by [hand] calculation bureaus in a reasonable amount of time. All types of mechanical calculators are also inadequate due to their inherent inaccuracy. Implementation of high-performance digital computers to solve such large problems would save a lot of time, materials and labor, plus it would require a relatively small staff of highly qualified specialists, whose only job would be formulating the task and evaluating the results. These factors dictate the necessity of constructing and putting into operation – as soon as possible – one or more high-performance digital computers designed to address the needs of the most important scientific centers. Besides general-purpose computers, it is critical to develop specialized machines for solving ballistics, weather forecasting problems, and so on. Finally, some very significant problems require machines with many of the elements (counting, programming) used in digital computers. This would improve the problem solving methods and allow us to achieve positive results faster and more frequently than we do today. The automatic digital computer, which is briefly described below, is based on an original design. The diagrams of computer elements – adder, multiplier, divider, interpolator, transformers from decimal to binary system and vice versa, plus a series of relay schemes – are presented here for the first time as far as we know. An objective comparison with the computers that are being built abroad (according to our information), shows that our proposed machine has major advantages, as described below. We present here the general design of the machine and its elements; it will require a detailed project plan and a great deal of experimental work on separate units before we can begin building and assembling the computer. II. General Description of the ATsVM The ATsVM is a general purpose computer. 1. Calculations are done automatically. The operator's involvement stops once he has prepared the machine to solve a problem. 2. The calculations are carried out in electrical relay-code circuits. Mechanical moving parts are used in a few of machine elements: the program unit, output printing device and several others. 3. The calculations proceed very rapidly: the machine is capable of performing at up to 2000 operations per second. 4. The computer is "digital." Calculations are carried out as numerical operations. The initial data and results are presented as ten-digit numbers in decimal form. Internal calculations are carried out in binary form. As the ATsVM project's basis, the high-performance digital computing machine had to satisfy the following requirements: 1. The machine must have units for executing the basic numerical operations: addition, subtraction, multiplication and division. Depending on the general scheme of the computer, there could be separate units for each operation or one calculating unit for all operations, since the addition unit could perform the subtraction operation using a supplementary number; multiplication could be done by sequential addition and division could be carried out by sequential subtraction. But using a separate unit for each operation increases the calculation speed and reduces the amount of memory required 2. In order to guarantee high performance in automatic mode, the machine needs a storage device, or memory, for internal data and calculation results. This memory device should receive and send data at the same speed as the numerical operations are performed, which in the electronic computer might be on the order of tens of microseconds. 3. The computer needs to have a table for number input. Reading the table may be done by the main computer or by a separate interpolator. The use of a separate interpolator could increase the computer's operating speed, simplify the programming process and reduce the required memory capacity. 4. A high-performance digital computer needs a control unit to manage calculations when solving specific sets of problems. The control speed needs to be the same as the speed of arithmetic operations.. 5. The controller will have to choose between two or more sequences of actions (according to predetermined criteria) and execute operations as a result of that decision. There also needs to be a unit that determines the sign of any given number and compares two numbers. 6. The computer should have devices for input of data and output of calculation results. The input and output devices should operate at the same speed as the control unit. 7. Finally, the digital computer must have some means to "transport" the numbers and program signals between the different parts of the machine. The automatic digital computer consists of the following basic elements: 1. A numerical keyboard device for data input, and a program tape puncher that automatically converts from decimal to the binary system. 2. A main program controller for all of the machine"s work. This controller selects the appropriate computer elements that are required for any given operation and controls the sequence and type of calculations. 3. A sign identifier that compares two numbers; it allows the main program controller to simultaneously select between two or more sequential operations and execute them, depending on the result received from the identifier. 4. Two adders. 5. A multiplier. 6. A divider. 7. Storage for "saving" numerical data, intermediate calculations, etc. 8. An interpolator for automatic calculation of intermediate function values, which are presented in an assigned table for small numbers of discrete values of the argument. The interpolator consists of a unit that automatically fills a table. 9. An output device for writing the results to tape (in binary form). 10. A device for converting output results from the binary to the decimal system and for printing them on paper. 11. Digital and command buses for connection between machine elements and transmission of program signals. The unit-diagram for the ATsVM computer is presented in Figure 1: Figure 1. A General View of the ATsVM The program for a sequence of numerical operations is written on programming tape in this logical format: "from where," "to where," "what to do." This corresponds to the computer's numerical or differential method of problem solving. In order for the computer to work according to this scheme, all of its elements have a common structure of input and output circuits. All digits and numerical signs are transmitted simultaneously from one element of the computer to another. The whole computer is served by a single digital bus (33 lines for digits and one line for signs) connected through "logic elements" (gates) to digital inputs and outputs of all computer elements. Gate units are controlled by the main program controller; their selection is produced by program signal decoders, connected to the program bus throughout the entire computer. Each decoder is assigned a number, the binary form of which is a key for the given decoder. This way, if the "from where" strip of tape has the decoder key for exiting the multiplier, and the "to where" strip has the decoder key for entering adder No. 1, then the number must be transmitted from the multiplier to adder No.1. The "what to do" strip of the program tape describes the operation to be done (for example, receiving, sending, erasing, multiplying, etc.) Besides decoder numbers and command signals, the program tape consists of a start pulse on each line (for each step), which initiates the computer elements that perform calculations at each step and other steps of the program where needed. The "numbers" strip contains written entry data previously converted to binary code. The input device that prepares the programming tape is the interface between the human operator and the machine and can only work at low speeds. Therefore, it is separated from the high-speed computer. The tape is prepared before running the program. To minimize the difference in speed between the computer and the input device, several input preparation devices may be used simultaneously for several problems. The programming tape is virtually wear free, so it may be used repeatedly when solving similar tasks, but the input data must be copied. For multiple repetition of the same calculation, the programming tape can be glued together and used as a continuous loop. There is another method for numerical data input into the computer where the numbers are written not on the programming tape but on special "numerical" tape. In this case, the numerical data could be read by a small capacity storage device that is constantly refilled with the "numerical" tape after receiving a signal from the main program controller. This method would also apply for inputting data into a table. The programming tape, prepared in accordance with the above mentioned logic scheme, is then placed into the main program controller which "reads" it and according to its data selects the elements of the computer that carry out the requested operation. Thus, the tape controls the sequence and type of each operation. It is necessary to mention that although the computer has fully centralized control, the main program controller selects separate elements of the computer and sends the command to start operations. The operation inside the element is executed automatically and independently under the control of a local autonomous controller. For example: the main program controller selects the multiplier and gives the signal for multiplication. From that moment, the local program controller of the multiplier executes the consecutive addition of partial products as many times as the number in the multiplier, shifting the partial product each time one digit to the left. The independent calculation cycle for separate elements is finished at the beginning of the next step-interval (with the exception of the interpolator). During the same step-interval only one element of the computer can work (with the exception of the interpolator). The computer works in controlled intervals, the duration of which is set by the speed the programming tape. This way, the speed of the computer is easily regulated from the slowest to the fastest, and is defined by the speed with which the arithmetic operations are executed (up to 2000 step-intervals per second). In cases when it is necessary to change the solution path due to the sign or magnitude of the module of intermediate calculation results, the programming tape must contain two or more ways to solve the task. The "what to do" section of the tape clarifies when a given solution path should not be taken, for example: if the number is "equal to," "less than," or "greater than." The compared number and intermediate calculation result are sent to the sign identifier. Depending on the result of the identifier's output, the correct solution path will be selected. A separate element of the computer is used for interpolation and every arithmetic operation except subtraction. This significantly simplifies programming, increases the operating speed of the computer, and reduces the amount of storage required. Two adders are used in the computer, one of which may be used as interim storage for a series of summations. Numerical data and intermediate results are stored in table form. In order to recall a number from storage, two keys are written on the programming tape. One key identifies the column and the other one the row of the storage table where the number is originally written. Therefore, writing the number and then reading it from storage requires two step-intervals. As already mentioned above, the required storage capacity depends on the problem being solved. It should be noted that even a relatively large storage capacity might be insufficient for certain solutions, for example: solving algebraic equations with several hundred unknowns. For such tasks, the storage capacity required would have to be several hundred thousand numbers. If the objective is to calculate at the computer's maximum speed, then such storage capacity would be difficult to realize because of the complexity and high cost of computer construction. Therefore, for solutions requiring a large memory capacity, the computer would have to operate at a lower speeds using large capacity tape storage. It would function as following: the intermediate calculation results are written on tape exactly the same way as the calculation results in the output unit - in the order in which they are received. Then, they are entered into the computer using the above-mentioned second method of numerical data input - into a small capacity storage device, which is constantly refilled with numbers from the tape. The numbers are stored in the same order as they will be used in subsequent calculations. An important advantage for the digital computer is the potential to input the numerical data in table form, which would require a table reader and, if necessary, an interpolator. In the computer, tables can be constructed in two ways: a) The function is presented in the form of a series: f(a + h) = C0 + C1h + C2h2 + C3h3 + ... b) The argument and the corresponding coefficient values are entered into the table C0, C1, C2, C 3 . . . Cn Reading and interpolating the table is executed by a separate interpolator that is in fact a simplified digital computer with fixed programming, which operates in much the same way as the main computer. For a given interpolation formula, the interpolator's program does not change and is written not on tape, but directly on a magnetic drum that is permanently rotating at high speed. The overall scheme of the computer is rather complex. However, it consists of several simple schemes: binary counters, gate elements that use an "on-off" principle, triggers, etc. Gate elements are the most common in the scheme. If the gates are constructed using electronic vacuum tubes, then their total quantity will increase significantly, with "gate" tubes amounting to 70% of all tubes. Taking this into account, we recommend substituting gate tubes with simpler components, such as magnetic and rectifier circuits. Although the time constant of magnetic circuits is greater than electronic ones, magnetic circuits could be used in a number of places. We will not speculate at this time which specific components (magnetic or rectifying) are best to be used as gate devices, and but we are certain that most of the gates can be built with the new schemes. We will also note that replacing electronic tubes in gate devices significantly simplifies the construction, increased reliability, durability and performance of the computer. The most promising long-range development for gate schemes is the implementation of crystal diodes, or rectifiers. Unfortunately, the Soviet Union is not producing these components yet. However, there is no doubt that we will begin making them very soon, because crystal diodes have many applications in modern radio and radar technology. Crystal diodes' miniature size and suitability for very high frequencies, combined with the absence of hot cathode ray tubes that have a limited life cycle and expend enormous heat energy, allows for the construction of extremely compact and inexpensive calculating units, which can be used for both stationary and mobile devices. The latter feature is very important for military applications. * * * |
Remember, this document dates back to the mid-1948, and Lebedev had not yet started building the MESM. He would later write, "I began working on high-performance electronic counting machines at the end of 1948."
In the West, ten computers were already being developed in America, one in England, and one in France. Since the computers were developed primarily for military purposes, published materials about them were very brief. Most of these computers were based on electro-mechanical relays rather than on electronic vacuum tubes.
After carefully reading the above report, it becomes obvious that Brook and Rameev came very close to realizing the principle of stored memory programs – originally considered von Neumann's and Lebedev's idea. They achieved it by allowing the programs to be written into memory (on tape), placing the result on the same tape and then inputting it back into the machine for subsequent calculations. In other words, they provided the option to process instructions in the computer's arithmetic unit.
This is what Rameev remembered about these breakthroughs:
Working in Berg's Institute No. 108 was like attending a fine school. The knowledge I gained in electronics combined with almost twenty years of amateur radio experience and my tendency to invent things may explain why collaborating with Brook was so successful. Together, Brook and I discussed general ideas about the computer. Then I drafted circuit schemes and wrote explanations for his review. He occasionally made notes and corrections to the text (I still have some of the copies with his handwriting). I worked out of his office, in the main building of the Power Engineering Institute.
We also discussed how to implement this project and decided it was necessary to establish a special construction bureau. I worked for two weeks in the Lenin Library, studying the special literature on design of industrial enterprises and factories.[4] As a result, this document was born.
I cannot remember where or what I ate back then, but I remember the room I lived in, where the housekeeper stored potatoes. It had a large wood-burning stove that I heated up with thick volumes of Legislation of Tsarist Russia that I found there. From 1944, I leased a room – or sometimes a corner of a room – for two to four months at a time, in many different Moscow districts, changing places ten times. Nobody wanted to register my residence officially and landlords were afraid to rent a room to an unregistered person for an extended period of time. I kept all of my belongings in three paper bags, and I moved with them from one place to another. In 1952, SKB-245 gave me my own room in a communal apartment.
During their year of working together, Brook and Rameev prepared and sent to the State Inventions Committee over 50 invention applications for the various computer units. However, many of the applications were returned without confirmation or with a list of questions. There were no computer technology experts on the Committee yet, and the individual who reviewed the applications was an electric motor specialist. Meanwhile, the Committee gradually accepted the applications. In December of 1948 they submitted the invention application for the "The Automatic Digital Computer" and received notarized Invention Certificate No. 10475 on December 4, 1948 – the first computer invention document in the Soviet Union.
In the beginning of 1949, Brook presented the theory of a digital computer at a closed meeting of the Scientific Council of the Dzerzhinsky Artillery Academy, where he had been a member since 1947. In the presentation, he demonstrated the model of the diode matrix arithmetic unit that Rameev had designed.
In 1949, the Army unexpectedly drafted Rameev as a radio-location specialist and immediately flew him to the Russian Far East. However, the rush was unwarranted, because Rameev had to wait for an assignment for one and half months, before being sent to teach at a submarine school. Brook lobbied non-stop to have Rameev come back to Moscow, and contacted Bruevich and Parshin for help. In the end, Rameev returned to Moscow where he found a letter offering him a new post as the laboratory chief at SKB-245.
In order to accomplish this, Parshin had to sign a special document assuming personal responsibility for the "son of an enemy of the people," due to the secrecy of the project at SKB-245.
Borrowing from some of his and Brook's earlier ideas, Rameev began collaborating with Yuri Bazilevsky, the manager of the digital computer department at SKB-245. That marked the beginning of the Strela project, which was met with enormous enthusiasm. The SKB-245 chief, Mikhail Lesechko, dedicated all of his organizational abilities to the success of this project. His staff, in fierce competition with the BESM project, worked around the clock to achieve what was seemingly impossible. It only took workers a couple of nights to assemble cumbersome equipment for cooling the enormous premises where Strela's assembled units were being installed.
An SKB-245 employee and Strela project participant, Evgenia Tikhonovna Semyenova, recalled the work atmosphere of that time:
In March 1950 I had asked the personnel department at Moscow Energy Institute for a job at the Scientific Research Institute, or SRI-10. During graduate placement, I agreed to go to work at the prestigious "mail box."[5]
But instead I was sent to the SKB-245, which nobody had heard of. But I didn't complain, I just took the order and went. I got very lucky: first, I found myself in Rameev's lab, where I worked for five years. Everything I learned from him during those years has stayed with me throughout my life. Second, I got to work with the founder and manager of SKB-245, Mikhail Avksentievich Lesechko, who was certainly a very interesting person and a talented manager. I have never met another person like him since. Third, and most important, was the work. We were building one of the first digital electronic computers in our nation. During the first few months there we read American magazines that featured articles about computers; thank God that our managers had a large supply of them. Rameev would generate ideas and then we developed them ourselves. I would have never found such work at the SRI-10!
SKB-245 and the Schetmash were established at the SAM site around the end of 1949 or in early 1950. SKB-245 had several departments and since all work was top secret, department titles were replaced by numbers. But sometimes, we called them by their managers' names or even by themes.
The First Department, the Security Department – like the name suggests – protected our work and watched all of our comings and goings. They issued notebooks, which were sewn with thread, numbered, and sealed. Every morning they gave us our briefcases with those notebooks and collected them at the end of the workday.
The Second Department was responsible for analog computers and was managed by Roman Vasilievich Plotnikov. The guys from the Moscow Energy Institute worked in this section: Zhenya Glazov and Misha Yankin. We were close friends, so we always knew what was going on in their department. Vitenberg, Sulim, Gena Petrov and several others also worked there.
The Third Department, headed by Bazilevsky, was ours. We worked exclusively on developing the Strela computer. I will talk about our work later on.
The Fourth Department was managed by Ephraim Arumovich Gluzberg, although he was later replaced by Dmitri Alekseevich Zhuchkov. This department designed standard programs for the Strela and evaluated their operation.
The Fifth Department dealt with finances.
The Sixth Department was developing a differential analyzer and was managed by Alexander Bednyakov.
Later, other departments were added.
Our department had several laboratories. Rameev's lab designed Strela's arithmetic unit and the operational memory component. I designed the multiplier-divider unit, and Boris Zaitsev the addition-subtraction unit.
There was yet another lab managed by George Prokudaev, where Sasha Larionov, Larissa Dmitrieva and Maya Kotlyarevska worked. They were all from the Moscow Energy Institute, but joined us a year after the project began. Prokudaev's lab designed external memory devices using cathode ray tubes, but something was wrong with them. The tubes were extremely unreliable so Rameev and Lazarev started designing external memory on a magnetic drum. The first Strela models were made with external magnetic drum memory.
Trubnikov's lab developed Strela's other external units.
Even thought Rameev often did not get along with his superiors, he was gentle with his subordinates and spoke in a clam, low voice. I remember that Rameev and Bazilevsky had several disagreements during that period. It was pretty normal because there were many debates about the structure of the machine and its component base. For example, should they use vacuum tubes or relays? Rameev insisted that we make the computer with vacuum tubes. I still remember standing in front of the unit frame that contained twenty two hundred of these tubes, holding a P6 electronic tube in my hand. It was not a tiny tube, but a real P6, ten centimeter long. The multiplier's frame was about five meters wide and at least two and half meters high, or perhaps even more.
We worked very hard, often through the evenings and into the nights, especially when high-level officials were about to visit. They came from the Central Committee of the Communist Party of the Soviet Union, and from various other Ministries. We had to prepare ahead of time for such visits and on the days they actually came to the see the facilities, we hid our soldering irons. Rameev would always make a joke, "Here we are, sitting around again with clean necks!"[6]
We could work as late as we wanted, but if somebody was even three minutes late at the start of the day, he was reprimanded by the deputy director-administrator. If they were twenty minutes late, the case was referred to a disciplinary committee. At the entrance of the building was a clock where we were supposed to punch in with our time-cards. The overseer was a tough and ruthless woman who refused to give anyone a break if they were ever late.
Now I understand that we designed the computer in an unbelievably short time. Moreover, we designed and built every single component of the machine. We started around March of 1950 and by the end of 1951 the documentation was sent to SAM factory in Moscow. Before the end of 1952, the first models were ready for adjustment.
In 1953 the working model of the Strela was presented to the Stalin Prize Commission. At the same time, Lebedev revealed the BESM computer. The Prize went to SKB-245, because Strela appeared better prepared for industrial production and was cheaper. Our colleagues joked that the reason Strela was cheaper was because we hadn't been paid overtime.
For that time Strela's performance was ordinary, with a 2000 operation per second speed, 2048 words of storage, and 43-digit word length. It was a three-address computer.
By the time the Stalin Prize was awarded, I had already left SKB-245 to teach at the Moscow Energy Institute. But, during my lectures on impulse technology, I always relied on circuit calculation methods developed in the Strela project and I remembered Bashir Iskandarovich.
There was one thing about SKB-245 that bothered me during the entire five years that I worked there: the tight security. A soldier stood at the entrance and there was no going in or out of the building during working hours without a special permission form signed by a manager. Even in case of an emergency at home, the soldier would not let you go. We used to work evenings and weekends, but it still didn't matter!
And then, the First Department! God forbid if I forgot to return my secret case with all of the notebooks, drafts, or even tiny scraps of paper; a severe reprimand and an immediate inquiry in front of the entire laboratory team. What rubbish! More than once I worked until late in evening while my son and mother waited at home, awake. I would be riding on the Metro and suddenly it would hit me: "The oscilloscope! Did I turn it off? The briefcase? Oh Lord, I don't remember! Oh, yes, I returned it before dinner..."
Rameev acted as the deputy chief designer for the Strela. The SAM in Moscow eventually produced seven models – technically the first industrially produced Soviet electronic digital machines. They were installed at the Academy of Science's computer center, the Institute of Applied Mathematics, and in closed government ministry computer centers that were supporting the space, nuclear weapons, and energy complexes.
Rameev recalled an episode in 1954 when the first Strela had been delivered and installed at the Institute of Applied Mathematics. Keldysh and Lesechko used to frequent the laboratory and during this particular visit Strela performed some very impressive calculations for several nuclear physics problems, which prompted Keldysh to proclaim, "If we could make five to seven such computers, the Soviet Union would be in great shape!"
[3]Brook and Rameev are referring to the Harvard Mark-1.
[4]: Known today as the Russian State Library, The former Lenin Library [in Russian: Biblioteka imeni Lenina] in Moscow contains some 40 million books and periodicals.
[5] Translator's Note:The phrase "mail box" designated secret Soviet defense enterprises such as weapons research institutes, production plants, and military divisions.
[6] Translator's Note: This refers to a well-known Soviet joke about getting ready for guests: A mother tells her son, "Your grandparents are coming soon. Please go and wash your dirty neck now." The son replies, "But what if they don't come? Then I would sit around like a fool with a clean neck!"