Cost evaluation of three ground-logging methods in a mixed broadleaved mountainous forest


Aysan Badraghi

Technical University Dresden, Faculty of Environmental Sciences, Dept. of Forest Technology, Dresdner Straße 24, 01737 Tharandt, Germany.

Jörn Erler

Technical University Dresden, Faculty of Environmental Sciences, Dept. of Forest Technology, Dresdner Straße 24, 01737 Tharandt, Germany.

Seyed A. O. Hosseini

University of Tehran, Faculty of Natural Resource, Dept. of Forestry and Forest Economics, Karadj, Iran.

Robert Lang

Technical University Dresden, Faculty of Environmental Sciences, Chair of Biodiversity and Nature Conservation, 01737 Tharandt, Germany.



Aim of study: To compare cost and productivity of three ground-based logging methods by skidder: 1, tree length method (TLM), 2, long length method (LLM) and 3, short length method (SLM).

Area of study: A mixed broadleaved mountainous forest stand in the Hyrcanian forests in northern Iran.

Material and methods: To develop time prediction models, all measurements of time were replaced by their decadic logarithms, and on the basis of the developed models, we simulated cost of 11 skidding turns depending on the diameter of the log (DL), skidding distance (SD), and the winching distance (WD) for TLM, LLM, and SLM.

Main results: Our results demonstrated that on average the net costs per extraction of one cubic meter wood (m3) were 3.06, 5.69, and 6.81 €/m3 in TLM, LLM, and SLM, respectively, and the most economical alternative depending on DL, SD and WD was a TLM. Furthermore, the results of simulated models suggest that as long as the diameter of the felled trees is less than 40 cm, the cut-to-length system is not an economical alternative. The cut-to-length method can be applied for trees with larger diameter (more than 40 cm), and in short skidding distance SLM is preferable to LLM but in cases of long skidding distance, LLM is more economical than SLM.

Research highlights: DLand SD were the main causes which influenced the productivity and cost of ground-based logging methods.

Additional Keywords: tree length method; long length method; short length method; skidding unit cost; logarithm base 10.

Abbreviations used: DL (diameter of the log); DT (delay time of time consumption); LL (length of logs); LLM (long length method); NL (number of logs); SD (skidding distance); SL (slope gradient in the loaded direction); SLM (short length method); TLM (tree length method); WD (winching distance).

Authors' contributions: Conceived, designed and performed the experimental work: AB, JE, SAOH. Analyzed the data: AB, JE. Developed the concept and wrote the manuscript: AB, JE, RL. All authors read and approved the final manuscript.

Citation: Badraghi, A.; Erler, J.; Hosseini, S. A. O.; Lang, R. (2018). Cost evaluation of three ground-logging methods in a mixed broadleaved mountainous forest. Forest Systems, Volume 27, Issue 2, e013.

Received: 29 May 2018. Accepted: 05 Sep 2018.

Copyright © 2018 INIA. This is an open access article distributed under the terms of the Creative Commons Attribution 4.0 International (CC-by 4.0) License.

Funding: The authors received no specific funding for this work.

Competing interests: The authors have declared that no competing interests exist.

Correspondence should be addressed to Aysan Badraghi:





Material and methods






Nowadays, due to technological evolution, evalua­­tion of machine productivity and calculation of the average cost of skidding for each hour and each cubic meter is necessary to compare their efficiency (Demir, 2010). Measuring productivity and costs of forest equipment and forest harvesting alternatives is an important aspect of an forest industry in order to increase the efficiency and decrease the operating costs. To do so, the harvest planner must understand the cost characteristics of the available logging systems under the influence of parameters such as the tree size (volume), skidding distance, winching distance, skidding direction or ground slope and number of logs (McDonagh, 2002; Nurminen et al., 2006; Jirousek et al., 2007; Eghtesadi, 2008; Jourgholami et al., 2008; Naghdi et al., 2010; Barari et al., 2011; Jourgholami & Majnounian, 2011; Ghaffariyan et al., 2012; Gilanipoor et al., 2012; Ghaffariyan, 2013). Moreover, quantifying these fac­tors is not sufficient to evaluate the productivity and efficiency effects. Mechanized harvesting operations are popular in Australia because of their productivity and efficiency, improved worker safety and the reduced cost of operations (Alam et al., 2012).

An important research tool which is common to compare the productivity of forest harvesting systems across varying conditions is time studies (Harstela, 1991; McDonald & Fulton, 2005; Nuutinen et al., 2008; Ghaffaryan et al., 2012; Mousavi & Nikooy, 2014). During the last decade, several studies evaluated the unit cost of ground-based cable skidders by used time study techniques. Their obtained unit net cost was as following: Parsakhoo et al. (2009), 1.32 €/m3; Jourgholami et al. (2008), 4.20 €/m³; Naghdi et al. (2010), 4.03 €/m3; Gilanipoor et al. (2012), 3.70 €/m3; Mousavi et al. (2012), 12.3 €/m3 (13.9 $/m3); Lotfalian et al. (2011), 4.70 €/m3; Barari et al. (2011), 7.29 €/m3; Çaliskan (2012), 0.98 €/m3 (1.1 $/m3); Hejazian et al. (2013), 4.12 €/m3 and 3.3 €/m3. Also, Naghdi & Bagheri (2007) and Ghaffariyan et al. (2012) found that in the ground-based cable skidder system, skidding phase in the cut-to-length method was more expensive than the tree length method, and Mousavi (2008) reported that LLM was more economical than SLM.

In a comparison of whole tree harvesting sys­tem cut-to-length involves the high initial cost of investment, repair and maintenance of the machine’s complex computerized system, and the inability of the felling-processing machine to handle stems with stump diameter larger than 60 cm. However, whole tree harvesting requires more woods workers, supervision, and support than cut-to-length harvesting. In addition, the hourly machine rate for the whole tree method system has also been higher than for the cut-to-length (LeDoux & Huyler, 2001). Comparison productivity and costs of a skidder system (whole tree) and a forwarder system (cut-to-length) in southern Alabama showed that the productivity was limited by the woods transport vehicles, and weekly production rates were 261 cords for the skidder system and 249 cords for the forwarder system (Lanford & Stokes, 1996). A study in the eastern part of Canada showed that the wood production cost per cubic meter in assortment logging method was higher than full tree logging method (Favreau & Gingras, 1998). In the research area of Akay (1998) cost for cut-to-length system by using four sawyers and a forwarder was 9.96 €/m3 and 12.4 €/m3 in the Black Sea and Aegean regions, respectively, and cost for whole tree system by using four sawyers, a grapple skidder, and a loader was 8.2 €/m3 in the Mediterranean region. Also, Adebayo (2006) found that the hourly machine rate for the whole tree method system was slightly higher than the cut-to-length harvesting system.

The main objective of this investigation is com­parison cost and productivity of three logging alter­natives (1- TLM, 2- LLMand 3-short SLM) depending on the DL, SD, and WD in ground-based cable skidders system in a mixed broadleaved mountainous forest.

Material and methodsTop

Site description

This study was conducted in the compartments 46, 107 and 41 of Jo Jadeh district in the Wood Industry of Farim forests (a part of the Hyrcanian forests). Some characteristics of study compartments and harvesting alternatives in case study can be seen in Table 1. The elevation is approximately 445–2,250 m a.s.l (i.e., above sea level) with a north and northwestern aspect. The original vegetation of this area is an uneven-aged mixed forest dominated by Fagus orientalis and Carpinus betulus, with the companion species Alnus subcordata, Acer platanoides, Acer cappadocicum, Ulmus glabra and Tilia rubra. The soil type is forest brown soil and the soil texture varies between clay-loam and silty-clay. The average annual rainfall recorded at the closest national weather station was 845.5 mm and the mean annual temperature is 11.5°C. The silvicultural system was applied as a combination of group- and single-tree selection. The total volume of primary transportation was carried out by skidders to landing areas that were prepared at the border of the road in the lower part of compartments. Timberjack 450C rubber-tired skidder used in this study was a normal articulated, four-wheel-drive vehicle weighing 10.3 ton (55% on the front and 45% on the rear axle) with an engine power of 177 hp (132 kW) and engine model of 6BTA5.9. It is equipped with a blade for light pushing of obstacles and stacking of logs. The skidder was fitted with tires the size of 24.5–32 inflated to 220 kPa on both front and rear axles, it had a ground clearance of approximately 0.6 m. Timber bunching was carried out by the winch that was installed in the near part of the skidder from the stump to the skidder and one end of the dragged round wood was in touch with the ground (Jourgholami & Majnounian, 2013). High, length and width of the machine was 3 m, 6.28 m and 3.1 m, respectively.

Table 1. Some characteristics of study compartments.

Data collection

To estimate productivity and cost a continuous time-study technique was applied. Time for each work element and accumulated time were measured by a deci-minute stopwatch in minutes and seconds. A work cycle and time elements were same for TLM, LLM and SLM which broken down in seven phases (traveling unloaded, releasing, hooking, winching, traveling loaded, unhooking, pilling and delays). For each trip, important variables of the time consumption such as the size of logs (DL and length (LL) of logs), the number of logs (NL), terrain slope gradient in the loaded direction (SL), SD and WD were recorded.

Data analysis to estimate productivity rates and cost

Normally scientific who use time study in forest harvesting operations predict a time consumption model by using the original times which are gathered by time studies in the field. But in this paper, all time study numerical data were transformed to their logarithmic data based on 10 and the time consumption model was developed on the basis of logarithmic time data.

Why do we use logarithms?

Logarithm modeling is indeed the most reliable method for predicting the consumption of woodworking time. Today’s concept of logarithms might make it seem strange that logarithms were really developed out of comparing velocities of arithmetically and geometrically moving points (Villarreal-Calderon, 2008). There are two reasons for this unusual and slightly more difficult method: (i) Specialists in mathematical statistics know that logarithms often fit better to time studies than numerical data (Sachs, 1984). (ii) Numerical data produces unsolvable contradictions and problems that do not occur with logarithmic data (Erler, 1984, 2012), e.g., in time study numerical data, arithmetic mean and performance mean are different but in the logarithm of data, both performances mean and arithmetic mean are congruent. For example, to do a task by two employees, one employee needs 5 min and his/her colleague needs 15 min (their arithmetic mean is 10 min). The fast worker’s performance is 12 times/h while his slower colleague’s performance is only 4 times/h (their performance mean is 8 times/h). If this performance mean is recalculated to the arithmetic mean, it would be 7.5 min. This result highlights the two different arithmetic means to do a task, 10 min as the average time consumption and 7.5 min as the time for average performance, which can cause a lot of problems in the interpretation of the statistics (Erler, 2012). To solve this problem we present a solution, as the first, when all collected numerical data of the time study are transformed to their logarithm data based on 10. After calculating statistics parameter for the logarithmic data, they are recalculated into numerical figures again. The recalculated mean in our example is located between the two means (7.5 min < recalculated mean <10 min). Therefore, in this research we transformed all the time study data to their logarithms on the base 10, and using logarithmic data we developed three logarithmic time prediction models. Finally, by the used logarithmic models, we simulated the net cost for 11 skidding cycles depending on DL, SD, and WD in TLM, LLM and SLM. In order to compare the different logging methods under comparable conditions, the cost of 11 skidding turns was simulated by developed model under maximum, mean and minimum of the influence parameters such as NL, LL, DL, SD, WD, SL. Except LL value, maximum, mean and minimum of all influence parameter values were same for all alternatives which were calculated by used recorded data in TLM, LLM, and SLM. These parameters are depending on the terrain, and it is possible to harmonize them for all alternatives, but the length of log depends on the harvesting alternative and that makes the relevant difference between the harvesting alternatives (TLM, LLM, and SLM). Therefore, the value of LL has to be held different in the alternatives.

Prediction time consumption model and cost

The logarithmic time prediction models were developed on the basis of linear regression in SPSS software (SPSS, Tulsa, USA, Version 19, 2011). But only influencing factors with significant correlation coefficients at the level of significance a = 0.05 were used to develop the logarithm models. Subsequently, by used logarithmic models, we simulated unit cost for 11 skidding cycles depending on DL, SD, and WD in TLM, LLM, and SLM. In order to estimate cost depending on DL, DL, NL, and LL were considered as influencing variables. The value of DL and NL were variable between 11 cycles and LL was variable between the alternatives (TLM, LLM, and SLM).However, other parameter values (such as SL, SD and WD) were same between 11 skidding turns which was set to the mean value. To estimate the cost depending on SD, SD and LL were considered as influencing parameters, and the value of SD was variable between cycles and LL was variable between alternatives, and the other parameters were set to the mean value. In order to estimate the cost depending on WD, WD and LL were considered as influencing parameters, which variable parameter between cycles was WD and LL was variable parameter between alternatives and the other parameters were set to the mean value in all 11 skidding turns in TLM, LLM, and SLM. Delay time of time consumption (DT) by prediction model in TLM, LLM, and SLM was measured as following (Eq. [1]):


Statistical arithmetic mean and logarithmical mean

The estimated time by the logarithmic models and the measured time on the field (real time) by time study techniques in TLM, LLM, and SLM are shown in Table 2.

Table 2. Statistical parameter values to compare arithmetic mean of numeric and logarithm data, and geometric mean.

The ratio between the estimated time by logarithmic models and the measured time by the time study was 93% for TLM, 91% for LLM and 92% for SLM, as well as, values of geometric means are less than the time study numeric data (Table 2).

Logarithmic time prediction model for simulate 11 skidding cycles

TLM prediction model (Eq. [2]):

LLM prediction model (Eq. [3]):

SLM prediction model (Eq. [4]):

Y is time consumption by model (min), LD is loading distance, TU is traveling unloaded, R is releasing, H is hooking, W is winching, TL is traveling loaded, UH is unhooking, P is pilling.

The most important affected factors in TLM were SD, WD, NL, and DL (Eq. [2]). In both cases of LLM and SLM the most important factors were SD, NL, WD, DL and SL (Eqs. [3] and [4]).

Unit net cost depending on DL, SD, and WD

The simulated cost models show that the mean net cost depending on DL (volume), SD and WD were 3.06, 5.69 and 6.81 €/m3, respectively. In TLM the maximum net cost was 3.23 €/m3, mean 3.06 €/m3 and the minimum was 2.94 €/m3; in LLM maximum cost reached 5.76 €/m3, mean cost 5.69 €/m3 and minimum of cost 5.63 €/m3. And in SLM the queue was 6.97 €/m3, 6.81 €/m3 and 6.67 €/m3 depending on the winching distance (Fig. 1 and Table 3).

Figure 1. Net cost/m3 depending on SD, skidding distance (a), DL, diameter of the log (b) and WD, winching distance (c) in TLM, LLM, and SLM.

Table 3. Gross and net effective costs depend on WD (winching distance), SD (skidding distance) and DL (diameter of the log), in TLM, LLM, and SLM.

Delay time in the prediction model

On average, 11, 12 and 15 % of the gross effective time of the time study was DT in TLM, LLM, and SLM, respectively. As mentioned in the methodological part, the DT to the prediction model was estimated by using the average of DT in time study in TLM, LLM, and SLM: DT = (11+12+15)/3 = 12.7%.


According to our assumption all values of estimated time (geometric means) by logarithmic formulas were lower than the time study numerical data (lying between two means). Also, estimated time consumption by the developed models and real time were at a close ratio, on average it was more than 91% (Table 2).

Knowing the factors affecting the production and costs of skidding has an important role in planning and organizing consumed budgeting, as well as, arranging expenditures to raise profitability (Hejazian et al., 2013). Overall, after analysing the data and applying the skidding time prediction models, it became clear that the time consumption was a?ected mainly by the skidding distance, the winching distance, the number of logs, the diameter and length of log and the slope, which was in line with previous studies such as McDonagh, 2002; Jirousek et al., 2007; Eghtesadi, 2008; Jourgholami et al., 2008; Naghdi et al., 2010; Barari et al., 2011; Jourgholami & Majnounian, 2011; Ghaffariyan et al., 2012; Gilanipoor et al., 2012; Ghaffariyan, 2013. On the other hand, mechanized harvesting operations are popular in Australia because of their productivity and efficiency, improved worker safety and the reduced cost of operations (Alam et al., 2012).

On average, the net cost for extraction of 1-m3 of wood was 3.06, 5.69 and 6.81 €/m3 in TLM, LLM, and SLM. These results are consistent with former studies about cost of ground-based skidders system (such as Jourgholami et al., 2008; Parsakhoo et al., 2009; Naghdi et al., 2010; Lotfalian et al., 2011; Çaliskan, 2012; Gilanipoor et al., 2012; Mousavi et al., 2012; Hejazian et al., 2013). Furthermore, the results have presented that depending on DL, SD, and WD the most profitable alternative is TLM, also some researchers such as Favreau & Gingras (1998), Naghdi & Bagheri (2007) and Ghaffariyan et al. (2012) found that the skidding phase of the cut-to-length method was more expensive than tree length method, as well as Lanford & Stokes (1996) and Akay (1998) reported that production cost reduction with the alternatives that have the biggest dimension of wood. According to Adebayo (2006) hourly machine rate for the whole tree method system was slightly higher than the cut-to-length harvesting system. In the cut-to-length system, our result demonstrated that cost is dependent on DL, which in small diameter LLM is more economical than SLM but in the biggest diameters no significant differences can be seen between LLM and SLM. As Figure 1b has shown the difference between the minimum and maximum cost in LLM and SLM were significantly higher, which indicates that the diameter influences on the cost of LLM and SLM are higher than TLM. With regard to this, we can suggest that when the diameter of the felled tree is smaller than 40 cm, the most economical choice for the forest manager is the TLM, but when the diameter of the felled tree exceeds 40 cm they can also choose LLM or SLM, depending on other influencing factors. Also, LeDoux & Huyler (2001) reported that the cut-to-length system is unable to handle stems with a stump diameter larger than 60 cm.

The range of costs depending on skidding distances was significantly different in SLM (Fig. 1a). This leads to the suggestion that in short distances SLM is the best alternative, whilst for longer distances LLM or TLM is to be preferred. Furthermore, the result indicates that WD has no important influence on costs in all alternatives (Fig. 1c).


In time study, numerical data produces unsolvable contradictions and problems (Erler, 1984), hence researchers can benefit from logarithms to solve this problem. Despite the limited numbers of time study data (30, 31 and 41 turns in TLM, LLM, and SLM, respectively) we found a high ratio (91%) between numerical and logarithmic data.

The most important part of the harvesting operation is the logging operation. In this context, from an economical point of view, we could suggest suitable harvesting methods in ground-based cable skidders system in a mixed broadleaved mountainous forest. Our finding implies that TLM can apply for trees, whose diameters are less than 40 cm, if a tree’s diameter is bigger than 40 cm suitable alternative can be LLM absolutely in longer skidding distance but for trees with the biggest diameter in short skidding distances, suitable harvesting method can be SLM.


Adebayo AB, 2006. Productivity and cost of cut-to-length and whole-tree harvesting in a mixed- conifer stand. Master thesis. University of Idaho, Moscow, USA.

Akay A, 1998. Estimating machine rates and production for selected forest harvesting machines operating in the western United States and determining the most economical machine combinations under representative conditions in Turkey. Master Thesis. Oregon State University, OR, USA.

Alam MM, Strandgard MN, Brown MW, Fox JC, 2012. Improving the productivity of mechanized harvesting systems using remote sensing. Aust Forestry 75 (4): 238-245.

Barari K, Sotoudeh Foumani B, Nobakht A, Moafi M, Ashrafi SB, 2011. Time study and efficiency (production and costs) of wheeled skidder HSM 904 in Caspian forests. World Appl Sci J 14 (5): 726-730.

Çaliskan E, 2012. Productivity and cost analysis of manual felling and skidding in Oriental spruce (Picea orientalis L.) forests. Ann For Res 55 (2): 297-308.

Demir M, 2010. Investigation of timber harvesting mechanization progress in Turkey. Afr J Biotechnol 9 (11): 1628-1634.

Eghtesadi A, 2008. Evaluation of wood productivity rate in primary and secondary transportation in Nekachub region. Iran J Forest Poplar Res 16 (2): 275- 291. [in Persian].

Erler J, 1984. Leistungsstreuung und Mittelwertsautomatik. Doctoral thesis, Göttingen University, Göttingen, Germany.

Erler J, 2012. Wissenschaftliches Arbeiten. Fiwa: Akademie, Reihe: Forstl. Technikmanagement, Band: Zumutbare Vorgabezeiten. Buchhandlung Findus, Tharandt, Germany. 50 pp.

Favreau J, Gingras F, 1998. An analysis of harvesting costs in eastern Canada. Special report NO. Forest Engineering Research Institute of Canada, SR-129.

Ghaffariyan MR, 2013. Comparing productivity-cost of roadside processing system and road side chipping system in Western Australia. J For Sci 59 (5): 204-210.

Ghaffariyan MR, Naghdi R, Ghajar I, Nikooy M, 2012. Time prediction models and cost evaluation of cut-to-length (CTL) harvesting method in a mountainous forest. Small-Scale For 12 (2): 181-192.

Gilanipoor N, Najafi A, Heshmatalvaezin SM, 2012. Productivity model and cost of steel tracked skidder LTT-100A in downward skidding (Case study: Research and educational forest of University of Tarbiat Modares). Iran J Forest 4 (3): 242-252. [in Persian].

Harstela P, 1991. Work studies in forestry. Silva Carelica 18: 1-41.

Hejazian M, Lotfalian L, Hosseini SAO, Fallah A, 2013. Analysis and evaluation of ground skidding costs of two stands with respect to production per hour and log quality. Bull Env Pharmacol Life Sci 2 (11): 24-29.

Jirousek R, Klvac R, Skoupý A, 2007. Productivity and costs of the mechanised cut-to-length wood harvesting system in clear-felling operations. J For Sci 53: 476-482.

Jourgholami M, Sobhani H, Etemad, 2008. Evaluation of the efficiency of timberjack 450 C wheeled skidder (case study: In Kheyroudkenar forest). Iran J Pajouhesh & Sazandegi 79: 28-36. [in Persian].

Jourgholami M, Majnounian B, 2011. Evaluating and comparison of two skidding methods using wheeled cable skidder (Case study: Kheyrud forest, Noshahr). Iran J Forest 3 (3): 189-200. [in Persian].

Jourgholami M, Majnounian B, 2013. Effects of rubber-tired cable skidder on soil compaction in Hyrcanian Forest. Croat J For Eng 34 (1): 123-135.

Lanford B, Stokes BJ, 1996. Comparison of two thinning systems, Part 2. Productivity and costs. Forest Prod J 46 (11/12): 47-53.

LeDoux CB, Huyler NH, 2001. Comparison of two cut-to-length harvesting systems operating in eastern hardwoods. Int J Forest Eng 12 (1): 53-59.

Lotfalian M, Moaf M, Sotoudeh B, Akbari RA, 2011.Time study and skidding capacity of the wheeled skidder Timberjack 450C. Int J Sci Techn Educ Res 2 (7): 120-124.

McDonagh K, 2002. Systems dynamics simulation to improve timber harvesting system management. Master thesis. State University Virginia, VI, USA.

McDonald TP, Fulton JP, 2005. Automated time study of skidders using global positioning system data. Comput Electron Agr 48: 19-37.

Mousavi R, 2008. Comparison of productivity, cost and environmental impact of two harvesting methods in Northern Iran: short-log and long-log. Doctoral thesis. University of Joensuu, Joensuu, Finland.

Mousavi SR, Nikooy M, 2014. Evaluation of tree forwarding by farm tractor in patch cutting of poplar plantations in Northern Iran. Small-Scale For 13 (4): 527-540.

Mousavi R, Nikooy M, Nekzad E, Ershadfar M, 2012. Evaluation of full tree skidding by HSM-904 skidder in patch cutting of aspen plantation in Northern Iran. J For Sci 58 (2): 79-87.

Naghdi R, Bagheri I, 2007. Evaluation of tree length and assortment logging methods with respect to timber production in Caspian Forest in the Northern Iran. Asian J Plant Sci 6 (1): 46-50.

Naghdi R, Firouzan AH, Nikooy M, Barari K, 2010. Evaluation of production and cost of HSM-904 skidder and timber jack C-405 in forests of wood and paper industries of Mazandaran. Iran J Nat Resour 63 (1): 91-102. [in Persian].

Nurminen T, Korpunen H, Uusitalo J, 2006. Time consumption analysis of the mechanized cut-to-length harvesting system. Silva Fenn 40 (2): 335-363.

Nuutinen Y, Väätäinen K, Heinonen J, Asikainen A, Röser D, 2008. The accuracy of manually recorded time study data for harvester operation shown via simulator screen. Silva Fenn 42 (1): 63-72.

Parsakhoo A, Hosseini SA, Lotfalian M, Jalilvand H, 2009. Efficiency and cost analysis of forestry machinery usage in Hyrcanian Forests of Iran. World Appl Sci J 6 (2): 227-233.

Sachs L, 1984. Applied statistics a handbook of techniques, 2nd edition. Springer-Verlag, NY, 716 pp.

Villarreal-Calderon V, 2008, A look at the history and uses of logarithms. Montana Mathematics Enthusiast 5 (2-3): 337-344.