ADA @1981? 'An Atlas of the Solar Ultraviolet Spectrum Between 2226 and 2992 Angstroms R. Tousey, E. F. Milone, J. D. Purcell, W. Palm Schneider, and S. G. Tilford Rocket Spectroscopy Branch Space Sciences Division 2% E 53» October 19 74 NAVAL RESEARCH LABORATORY Washington, D.C. Approved for public release; distribution unlimited. PREFACE This atlas was made possible by the invention of the echelle diffraction grating by Harrison of the Massachusetts Institute of Technology in 1949, followed soon after by the production of echelles jointly with Bausch and Lomb, Inc. under David Richardson. The idea of making use of an echelle grating in a rocket-home spectrograph to map the solar spectrum at high resolution from 2000 to 3000 A occurred to R. Tousey and F.S. Johnson in about 1952, only six years after the first spectra in this wavelength range were obtained by the Naval Research Laboratory. Work on the project began at once, and in 1961 and 1964 excellent echelle spectra were ob- tained. With the publication of this atlas, and the line list, the original goal has been achieved. The unusual format of the echelle spectra required the development of special procedures to derive accurate values of wavelength and intensity; the large amount of information dictated processing by computer. However, personal judgment was required at every step. Extraction of the information contained in all the exposures from both flights would prob- ably increase, slightly, the accuracy of the atlas. However, we feel reasonably comfortable with the results and believe that little would be gained by striving for absolute perfection. Many individuals have been involved in this program over the years. The paper in the ap- pendix makes proper acknowledgments for the experimental work ending with the flights of 1961 and 1964. The preparation of the intensity atlas was commenced by L.E. Giddings in 1966. His work was taken over in 1967 by E.F. Milone, then a staff member at Gettysburg College and now at the University of Calgary. Others to whom we are indebted are Eugene McClurken and LB. Fox, each of whom spent a summer on the project; Rocket Spectroscopy Branch personnel D.L. Garrett and J.-D.F. Bartoe, who participated in the early reduction stages; and R. Hedler, among other Upper Air Physics Branch personnel, who helped with the formats of the final data tapes. R. Tousey ii CONTENTS PREFACE .......................................................... ii ABSTRACT .......................................................... iv INTRODUCTION ..................................................... 1 DESCRIPTION OF THE ATLAS ........................................ 1 PROCEDURES FOR REDUCTION TO INTENSITIES ........................ 1 CRITIQUE OF THE RESULTS .......................................... 4 REFERENCES ...................................................... 4 APPENDIX - An Echelle Spectrograph for Middle Ultraviolet Solar Spectroscopy from Rockets ............................................ 5 ATLAS ............................................................ 9 iii ABSTRACT The photochemistry of the earth’s upper atmosphere is controlled fundamentally by the incident radiation from the sun, especially the ultraviolet that is absorbed at levels above the tropopause. This atlas presents the incident solar irradiance with the high spectral resolution necessary to calculate the reaction produced when this energy is absorbed by atmospheric mo- lecular gases. Many of these molecules have absorption bands that consist of very narrow lines. Solar irradiance in units of ,uW cm“2 A—1 are presented in the wavelength range of 2226 to 2992 A. The curves were determined from high-resolution rocket echelle spectrograms photo- graphed above the earth’s atmosphere. Each panel of the atlas covers 14 A, with some overlap from panel to panel. The spectral resolution is approximately 0.03 A. iv AN ATLAS OF THE SOLAR ULTRAVIOLET SPECTRUM BETWEEN 2226 AND 2992 ANGSTROMS INTRODUCTION This atlas presents the sun’s absolute spectral irradiance over the range 2226 to 2992 A with approximately 0.03-A spectral resolution. The spectra on which it is based were photographed on Eastman Kodak IV-O film on August 29, 1961 at altitudes of more than 350,000 ft, where the atmos- pheric attenuation is close to zero. Twelve exposures of dura- tion from 2 to 84 sec were made using the echelle grating spec- trograph described by Purcell, Garrett, and Tousey [1,2] in the appendix. The appendix describes the instrument in detail and gives the instrumental profile. The instrument had the unique advantages of packing a spectrum covering nearly 2000 A with 30-mA resolution onto an area of film approxi- mately 2.5 cm2 and a size and shape that allowed it to fit into the instrument section of an Aerobee rocket, a volume of about 1 ft3. The echelle spectrograph had several disadvantages. Most serious was the high intensity of background stray light, characteristic of the earliest echelle gratings. To minimize stray light, a double-monochromator predisperser designed by J. D. Purcell was used. A knife edge in this device was ad- justed to reject direct sunlight of wavelengths greater than 3300 A. Nevertheless, all exposures made during the 1961 flight were exposed to some residual scattered light. This be- came increasingly important at the shorter wavelengths, where the solar spectrum becomes weaker, and finally swamped the spectrum at wavelengths less than about 2280 A at order 107. On a subsequent flight in 1964 the predisperser was adjusted to reject wavelengths greater than 2500 A (see appendix.) As a result, spectra from that flight had a back- ground that was freer from stray light. However, because of not quite perfect adjustment, the spectral resolution was slightly lower than for the 1961 results. A second disadvantage of the echelle lay in the fact that the two-dimensional array of spectra contained distortions of sev- eral types. As a result, an extremely complicated procedure was required to measure accurately densities and wavelengths with the microdensitometer and to determine correct values of intensity by photographic photometry. Each spectral order was slightly curved and the tilt of the lines changed along each order. The direction of dispersion of the residual stray light was not quite parallel to that of the echelle, as explained in the appendix; this caused the accurate measurement of the background by microphotometry to be impossible. The trans- mittance of the instrument varied from the center to each edge of each order and also of course from order to order; moreover, the wavelength dispersion varied from order to Manuscript submitted June 21, 1974. “Ha. y.. k swflggsflv . ~ 3600 m .> ' .‘ 3400 2» we anal) an “hem ~ WAVELENGTH (ANGSTROMS) 2 200 Xma- 4—10 ANGSTROMS» Fig. 1 — The middle ultraviolet spectrum of the sun. This com- posite is from 12 echellograms on 35-min, Kodak IV-O film taken from an Aerobee rocket on August 29, 1961. Portions of the best films have been edited and assembled here. These data have been reduced to produce the intensity atlas. order and changed slightly across each order. Finally the separation between orders and height of each order decreased with increasing wavelengths. These properties can be seen in Fig. 1, a composite of prints made from several exposures, and in Fig. 10 of the appendix. Because of these characteristics, it was necessary to calibrate and standardize the echelle spectra by reference to existing well-calibrated solar spectra of lower spectral resolution. DESCRIPTION OF THE ATLAS The atlas curves present on a linear scale the solar irradiance above earth’s atmosphere at 1 AU. in units of uW cm'2 A1. The wavelength range extends from 2226 A at the beginning of order 109 to 2992 A at the end of order 82. Each sheet covers 14 A , providing some overlap from sheet to sheet. The atlas is arranged so that the beginning or the end of an order coincides with an edge of a sheet. The edges of orders are indicated by heavy broken horizontal lines, which show ap- proximately those regions where the intensity values and wave- lengths are somewhat less accurate. Each sheet of the atlas is based almost exclusively on the particular exposure from the 1961 spectra that was judged best for that order, considered from the points of view of density, contrast, freedom from stray light, and emulsion defects. Other exposures were used to check the procedures employed for conversion to intensities and to fill in where defects such as pinholes were present. These intervals are indicated by broken sections of the intensity curves. Because of the widespread interest in the Mg II H and K lines, an addi- tional sheet is included spanning the region 2794 to 2800 A containing intensities from exposure 9 of the 1961 flight film. As an independent overall check on the curve derived from the 1961 spectra, a comparison was made with several orders of the 1964 spectra, reduced and normalized in the same way. The accuracy of the values of irradiance is difficult to estimate. Except near the ends of orders and in regions where the curve is shown as broken, the absolute values of irradiance are be- lieved correct within i20%. The Fraunhofer line profiles in the atlas curves apply to the radiation emitted by a central region of the solar disk of ap- proximately 10 arcmin in diameter. The mean intensity level of the curve, however, is normalized to give the irradiance from the entire sun. Therefore, conversion to solar radiance, without taking limb darkening into consideration, will result in absolute values that apply to a mean over the sun’s disk. PROCEDURES FOR REDUCTION TO INTENSITIES A straightforward reduction of the spectra to intensities by means of H and D curves was not possible because the scattered-light background was not uniformly distributed over the film but varied in a complicated way with wavelength and order. The basic procedure employed was that described in the appendix, which proposed treating the stray light as a parameter because a meaningful value of stray light back- ground could not be determined directly from the echelle ex- posures. This parameter and the contrast of the H and D curve were to be varied until the intensity curves reduced from the echelle spectrum agreed with the intensity curve obtained from reference spectra that had approximately the same instru- mental resolution but were free from stray light. The pro- cedure was tested by Tousey et a1 (see the appendix), and satisfactory agreement was obtained between the echelle and the G'ottingen atlas [3] in the region 3330 to 3338 A , spectra that had approximately the same instrumental resolution. To apply this procedure over the wavelength range 2226 to 2992 A, however, required making comparisons with spectra having much lower resolution than the echelle. This was done by Gaussian smoothing the echelle spectra to match the com- parison spectra. The description of the detailed procedure follows. First, the echelle spectra were measured with a Jarrell-Ash microphotometer adjusted to read out the central strip along each order. The slit was moved transversely during scanning to keep it centered on the region of maximum density of the slightly curved order. The microphotometer output in trans- mittance was recorded on magnetic tape, converted to densi- ties using a CDC 3800 computer, and stored on cards. In the next step, the difficult task, the density data were converted to intensities and normalized against the reference spectra using a CDC 3100 computer. The reference spectra used were the rocket spectra of Wilson, Tousey, Purcell, Johnson, and Moore [4] covering the 2990 to 2635 A region and of Malitson, Purcell, Tousey, and Moore [5] covering the region 2635 to 2085 A. These spectra had been normalized to apply to the radiation from the entire sun by comparison with spectra of still lower resolution that were exposed to the radiation from the entire sun. Numerical values of the refer- ence spectra intensities had been derived from the published curves and were available in digital form from Brinkman, Green, and Barth [6] . The resolution of these reference spec- tra was approximately 0.6 A. Therefore the echelle data were subjected to an averaging and Gaussian smoothing process that deresolved them to approximately 0.6 A resolution. The comparison of the deresolved echelle spectra and the Naval Research Laboratory (NRL) reference spectra was carried out in short sections of each order because of the characteristics of the echelle, previously discussed in the In- troduction. An approximation to the background density was obtained from the strips of film above and below each order. The difference between densities of the deresolved spectra and the approximate background densities was fitted section by section to the reference spectra to achieve the best overall fit. For each section the fit was achieved using the following equation: logI=A(D —D0) +3 + Cx where x is the distance from the center of the order D is the density of the spectrum D0 is the mean of the two background densities adjoining the spectrum at each position I is the irradiance of the reference spectrum. The parameter A is the reciprocal of the effective contrast; parameter C adjusts for nonuniform instrument efficiency across an order, and parameter B serves to make the correction from order to order and to adjust to the absolute level of the reference spectrum. The need for these adjustments is appar- ent from Fig. 9 of the appendix. In this process a correction to the value of D0 is automatically introduced so that the optimum fit at the centers of Fraunhofer lines was obtained. The various wavelength ranges and orders involved in the fit- ting process are listed in Table 1. The final stage in the reduction was to apply this equation, using the values of fitting parameters determined as described, to the respective sections of the unsmoothed echelle density data. A discussion of the general operation, the original den- sitometer system, and some of the computer programs in- volved has been published by Ditzel and Giddings [7 , 8]. The curves were plotted by the computer; however, corrections by hand were required, to take care of artifacts. Table 1 also includes data concerned with the determina- tion of wavelengths. The wavelength scale was computed separately for each order using wavelength values from the echelle line list (published separately) and a least-squares third- order-polynomia] fitting procedure over the 28mm of each spectral order. The mean standard error e and the number of lines n used for making each fit are tabulated. The wave- length accuracies achieved were approximately 0.005 A. How- ever, wavelengths can be read from the curves to no better than 0.02 A. Table 1 Principal Sections of the 1961 Echelle Spectra Used in Preparation of the Atlas . Mean . Fitting Section Limits _ Mean . Fitting Section Limits Exposure Exqposure Altltude Order Standard Lines Used - ' . Exposure Exposure Altitude Order Standard Lines Used . . ‘ Number 1me Range Number Error Per Flt Beginning Ending Number T1me Range Number Error Per Flt Beginning Endmg (S) (km) (6) (n) Wavelength Wavelength (5) (km) (6) (n) Wavelength Wavelength (A) (A) (A) (A) 2 8 132-139 109 $00078 36 2226.0 2228.9 2 8 132-139 102 i0.0038 19 2376.0 2380.4 2228.9 2235.0 2380.4 2382.3 2235.0 2241.1 2382.3 2388.0 2241.1 2244.1 2388.0 2397.1 2244.1 2252.0 2397.1 2402.2 2402.2 2404.6 108 0.0039 14 2249.0 2253.2 2404.6 2406.0 2253.2 2258.1 2258.1 2264.8 101 0.0043 43 2399.0 2405.3 2264.8 2268.7 2405.3 2414.0 2268.7 2273.0 2414.0 2422.0 2422.0 2428.2 107 0.0042 23 2269.0 2275.6 2428.2 2430.0 2275.6 2277.6 2277.6 2279.7 100 0.0046 41 2423.0 2431.3 2279.7 2282.9 2431.3 2439.1 2282.9 2285.6 2439.1 2447.9 2285.6 2291.3 2447.9 2453.0 2291.3 2294.0 99 0.0037 7 2447.0 2462.7 106 0.0039 28 2286.0 2288.2 2462.7 2473.5 2288.2 2289.7 2473.5 2480.0 2289.7 2299.7 2299.7 2302.5 98 0.0037 21 2472.0 2477.4 2302.5 2305.9 2477.4 2491.5 2305.9 2310.6 2491.5 2505.0 2310.6 2313.0 97 0.0047 32 2497.0 2504.1 105 0.0063 28 2312.0 2317.9 2504.1 2510.5 2317.9 2322.9 2510.5 2524.7 2322.9 2328.6 2524.7 2529.0 2328.6 2337.0 96 0.0043 53 2524.0 2529.5 104 0.0018 21 2333.0 2338.0 2529.5 2549.2 2338.0 2344.4 2549.2 2557.0 2344.4 2347.6 2347.6 2350.2 4 4 141-144 95 0.0043 62 2550.0 2580.0 2350.2 2353.2 2580.0 2583.0 2353.2 2355.1 2355.1 2360.0 94 0.0036 31 2577.0 2584.0 2584.0 2605.3 103 0.0039 29 2352.0 2360.6 2605.3 2611.0 2360.6 2365.1 2365.1 2374.8 93 0.0044 36 2605.0 2617.1 2374.8 2376.8 2617.1 2629.1 2376.8 2382.0 2629.1 2634.4 2634.4 2639.0 92 0.0031 42 2633.0 2650.1 2650.1 2656.4 2656.4 2668.0 (Table Continues) Table 1 (Continued) Principal Sections of the 1961 Echelle Spectra Used in Preparation of the Atlas Mean Fitting Section Limits Mean Fitting Section Limits Exposure Altitude Lines Used Exposure Altitude Lines Used E35322? Time Range N235; Stggiird Per Fit Beginning Ending E13322: Time Range N01113:); Stggiird Per Fit Beginning Ending (5) (km) (6) (n) Wavelength Wavelength (5) (km) (e) (n) Wavelength Wavelength (A) (A) (A) (A) 9 2 176-175 91 10.0043 42 2662.0 2666.8 2826.4 2829.7 2666.8 2671.6 2829.7 2836.9 2671.6 2676.1 2836.9 2841.3 2676.1 2679.3 2841.3 2849.0 2679.3 2690.7 2849.0 2854.0 3233; 32333 85 0.0035 53 2850.0 2851.5 ' ' 2851.5 2853.6 90 0.0039 31 2691.0 2698.8 28535 ”591 2698.8 2704.3 2859.1 2868.9 2704-3 2707.6 2868.9 2872.2 2707.6 2712.1 2872.2 2879.3 27121 2715.0 2879.3 2883.5 2883.5 2888.0 2715.0 2720.6 272% 27260 12 13 124-110 84 40.0041 60 2884.0 2886.4 2886.4 2890.9 89 0.0047 35 2722.0 2726.1 2890.9 2894.0 2726.1 2744.9 2894.0 2901.3 2744.9 2753.3 2901.3 2906.5 2753.3 2755.1 2906.5 2911.4 2755.1 2757.0 2911.4 2913.6 1913.6 2919.7 88 0.0040 39 2753.0 2756.6 2919.7 2922.0 2756.6 2790.0 83 0.0044 60 2918.0 2923.0 87 0.0049 20 2794.0 2803.5 2923.0 2933.2 2803.5 2808.0 2933.2 2936.8 2936.8 2947.3 12 13 124-110 87 0.0050 38 2785.0 2792.6 2947.3 2950.0 2792.6 2803.2 2950.0 2953.5 2803.2 2811.3 2953.5 2958.0 2811.3 2821.0 82 0.0039 99 2954.0 2966.9 86 0.0035 53 2817.0 2819.7 2966.9 2981.6 2819.7 2826.4 2981.6 2992.0 CRITIQUE OF THE RESULTS The procedure as it applies to the determination of intensi- ties over a given section assumes that the H and D curve, ad- justed for stray-light density, contrast, and speed change, is linear; that is, the shoulder and toe are disregarded. Errors arising from this assumption were minimized by choosing for each order, to the greatest extent possible, the exposures whose density fell within the linear part of the H and D curve. However, deep Fraunhofer lines and regions of unusually high intensity are somewhat less accurate. The consistency of the results can be assessed by comparing spectral intensity plots derived from different exposures. Comparisons between exposures 12 and 9 at order 87 spanning the range 2784 to 2823 A , and between exposures 9 and 4 at order 91 for the range 2662 to 2699 A show substantial agree- ment across each order with the exception of regions near the edges Where the focus and the background light change rapidly. This is illustrated by Fig. 2 in which a 5—A portion of the latter range is reproduced. Curve 4 is displaced upward by 1.5 units from 9, which is taken from the atlas. It can be seen that the agreement is good, although some information was lost in ex- posure 4 because of its greater density. Another comparison can be made from the atlas itself, which includes the curve from exposure 9 from 27 94—2808 A, a shorter exposure than the principal atlas curve which was from exposure 12 in this region. It is of interest to compare the absolute irradiance values of this atlas with results obtained by others. The atlas pro- duced by McAllister [9] covers this region with approxi- mately 0.3-A resolution; however, it was normalized to the same reference spectra used in the calibration procedure for the echelle spectra. Completely independent data were obtained by Kacalov and Jakovleva [10] . Their spectra were obtained with a rocket-borne concave grating spectrograph. The results cover 3100 to 2700 A and 2700 to 2570 A with resolution values of 0.15 and 0.3 A, respectively. For a detailed discussion of these results see Tousey [2] . Regions from their curves selected at random were compared with the corresponding portions of the present atlas, which were deresolved to the same spectral resolution. The comparisons are shown in Fig. 3 through 7. The results agree as well as one could expect. Some of the differences probably result from the deresolution process. A rather critical test is the comparison of the range 2790 to 2810 A that includes the Mg 11 H and K lines. Figures 6 and 7 compare the curves of Kacalov and Jakovleva over the K and H lines, respectively, with the 1961 spectra deresolved to the same resolution. For Mg II the NRL curve for K1 was signifi- cantly lower, but for H1 it is nearly coincident. The agree- ment for H2, 3 and K2, 3 is satisfactory. To examine this further, the echelle spectrum from the 1964 flight was cali- brated and deresolved exactly as was done for the spectrum from the 1961 Flight. The curves over the Mg II K line, as compared in Fig. 8, are in good agreement. Intensity, l 6 I l l l ;, 4—91 _. I II I II I II A | ll ‘ .' II l. .' 4 Ii 5 -' l I' I , I l l [1' ,' " . if” [\I‘ , llll I I' l 2— II I I I - I/\ A I l ‘I' f I 9—9l l ..... . l.. .I ....... I... I. 11.. 2692 Wavelength (A) 2695 Fig. 2 — Intensities from the 9lst order of the 1961 echelle in the 2691 to 2696 A region from two different exposures. Exposure 4 is offset above exposure 9 by 1.5 units of irradiance (,uW cm‘2 A“1). Exposure 4 has a much higher density in this region, causing some loss of information. Exposure 9 was used for the atlas. 0.5 ——NRL -—- K&J 2500 2510 2520 Wavelength (A) Fig. 3 — A comparison between the 97th order of the echelle atlas (solid line) and the Kaéalov and Jakovleva [11] atlas, (dashed line), in the 2495 to 2520 A re— gion. The echelle atlas segment has been smoothed with a Gaussian function to approximate the resolution of the Kacalov and Jakovleva curve. The vertical axis is in units of the logarithm of solar irradiance (MW cm—2 A*1). l.0 ‘2723\’ 2726 v 2729 Wavelength (A) Fig. 4 —- A comparison between a smoothed portion of the 89th order of the echelle atlas (solid line) and the Kac’alov and Jakovleva [11] curve (dashed line), in the 2723 to 2730 A region. (Units as in Fig. 3.) ———NRL ———K8.J 2735' Wavelength (A) Fig. 5 - A comparison between a smoothed portion of the 89th order of the echelle atlas (solid line) and the Kaéalov and Jakov- leva [11] curve (dashed line), in the 2730 to 2740 A region. (Units as in Fig. 3.) 0.5F —NR|_ log I o.o~ "’\ l l l l l I l l I 2795 2799 Wavelength (A) -0. I 5 2791 Fig. 6 — A comparison between a smoothed portion of exposure 12 at the 87th order of the echelle atlas (solid line) and the Ka- calov and Jakovleva [11] curve (dashed line) in the 2790 to 2800 A region. (Units as in Fig. 3.) 0‘5F ___NRL —~-K&J logI 0.0 .05 I I I I l l I I J 2805 2809 Wavelength (A) Fig. 7 —- A comparison between a smoothed portion of exposure 12 at the 97th order of the echelle atlas (solid line) and a panel of the Kaéalov and Jakovleva [11] atlas (dashed line), in the 2800 to 2810 A region. (Units as in Fig. 3.) 0.5 _ '9“ Fig. 8 — A comparison of the "' '9“ 87th order of exposure 12 of the August 1961 flight and a similarly treated portion from exposure 1 of the 1964 flight film. (Units as in Fig. 3.) Both spectra have been Gaussian smoothed to the same resolution as the Kacalov and Jakovleva spectra. The y axis scale is the same as the one used in Figs. 4 and 5. log I —0.5 ‘2794 2796 Wavelength (A) Additional results over the region of the Mg II H and K lines have been reported by Lemaire [11, 12] and Bates et a1. [13] . Their spectra were of very high resolution and were also stigmatic. They showed that the profiles of these lines vary greatly with activity and also with position relative to the limb. Since the compilation of this atlas, new measurements of the solar irradiance have been made by Broadfoot [14] using an Ebert-Fastie spectrometer flown in an Aerobee—150 rocket. The spectral resolution was 3 A half width. According to Broadfoot, the intensities of the N RL reference spectra that had 0.6 A resolution were in excellent agreement with his own results over the entire range covered by this atlas, with the exception of the regions 2500 to 2600 A and 2850 to 3000 A. There the values of Broadfoot were lower than those of NRL by as much as 25%. No reason for this difference is apparent, and these independent results agree to well Within the stated accuracies. Therefore, correction to match Broad- foot’s results is regarded as optional. There is, of course, the question whether the sun’s spectrum is affected by solar activity and the solar cycle in the spectral range covered by this atlas. Here the radiation from the center of the disk is almost entirely photospheric. Exceptions are the Mg II emission cores and possibly a few emission lines such as Si II at 2334.50 and 2350.28 A. Although the solar constant has not been shown to vary with solar activity, there remains the possibility that the contribution from the short wavelength end of the photospheric spectrum may vary slightly with the solar cycle. REFERENCES 1. J.D. Purcell, D.L. Garrett, and R. Tousey in Space Research 111 (W. Priester, editor), North-Holland Publishing 00., Amsterdam, 1963, p. 781-786. 2. R. Tousey, Space Sci. Rev. 2, 3 (1963). 3. G. Bruckner, Photometric Atlas of the Near Ultraviolet Solar Spectrum, Vanderhoeck and Reprecht, Gottingen, 1960. 4. N.L. Wilson, R. Tousey, J.D. Purcell, F.S. Johnson and CE. Moore, Astrophys. J. 119, 590 (1954). 5. H.H. Malitson, J.D. Purcell, R. Tousey, and C.E. Moore, Astrophys. J. 132, 746 (1960). 6. RT. Brinkman, A.E.S. Green, and C.A. Barth, JPL Technical Re- port 32-951, Jet Propulsion Laboratory, Pasadena, Calif., 1966. 7. E.F. Ditzel and LE. Giddings, Jr., Appl. Opt. 6, 2085 (1967). 8. ER Ditzel and LE. Giddings, Jr., “A Computer Method for the Automatic Reduction of Spectroscopic Data,” NRL Report 6632, Jan. 1968. AD 665281. 9. H.C. McAllister, A Preliminary Photometric Atlas of the Solar Ultraviolet Spectrum from 1800 to 2965 Angstroms, University of Colorado Printing Services, Boulder, 1960. 10. V.P. Kachalov and A.V. Jakovleva, Izv. Krymskyi Astrofiz. Obs., Akad. Nauk. SSSR 27, 5 (1962) (NRL Translation 923). 11. P. Lemaire, Astrophys. Ltrs. 3, 43-48 (1969). 12. P. Lemaire, in Ultraviolet Spectra and Related Ground-Based Observations, (L. Houziaux and HE. Butler, editors), Reidel, Dordrecht, Holland, 1970, p. 332. 13. B. Bates, D.J. Bradley, C.D. McKeith, N.E. McKeith, W.M. Burton, H.J.B. Paxton, D.B. Shenton, and R. Wilson in Ultra- violet Stellar Spectra and Related Ground—Based Observations (L. Houziaux and HE. Butler, editors), Reidel, Dordrecht, Holland, 1970, p. 274. 14. A.L. Broadfoot, Astrophys. J. 173, 681 (1972). rinted from APPLIED OPTICS, Vol. 6, pjage 365. March 1967 e Re Copyright 1967 by the Bptical Society of America and reprint y permission of the copyright owner An Echelle Spectrograph for Middle Ultraviolet Solar Spectroscopy from Rockets R. Tousey, J. D. Purcell, and D. L. Garrett An echelle grating spectrograph is ideal for use in a rocket when high resolution is required because it occupies, a minimum of space. The instrument described covers the range 4000—2000 A with a resolution of 0.03 A. It was designed to fit into the solar biaxial pointing-control section of an Aerobee-150 rocket. The characteristics of the spectrograph are illustrated with laboratory spectra of iron and carbon arc sources and with solar spectra obtained during rocket flights in 1961 and 1964. Problems encountered in analyzing the spectra are discussed. The most difficult design problem was the elimination of stray light when used with the sun. Of the several methods investigated, the most effective was a predis- persing system in the form of a zero—dispersion double monochromator. This was made compact by folding the beam four times. I. Introduction High resolution spectroscopy from rockets requires the design of extremely specialized equipment, since it is hardly possible to fly a 21—ft (6.4 In) concave grating spectrograph in the solar—pointed section of an Aerobee rocket with length but 3 ft (0.9 m) and cross section 6 in. X 8 in. (15 cm X 20 cm). Even now, after twenty years of research with rockets, this is the largest space available in this country for solar instrumentation that must be recovered after flight. The design of a high resolution spectrograph, covering the wavelength range 4000 A to 2000 A, approximately, that was compact enough to fit into this space was made possible by the de- velopment of the echelle diffraction grating by Harri— son.1 If produced by a conventional spectrograph, this spectrum would have been 5 ft (2.3 m) in length, nearly twice as long as the instrument space. When cut into sections and stacked by means of the echelle and prism combination, the entire spectrum covered an area only 1 in. (2.5 cm) on a side. Thus, the echelle system was ideally suited for high resolution rocket spectroscopy over a wide wavelength range. The original design of the echelle spectrograph was undertaken in 1952 by Tousey and Johnson, and was completed by Purcell. The first solar spectra were ob— tained in 1956 by Purcell et al.2 Greatly improved spectra were photographed on 29 August 1961 and 12 November 1964 by Purcell et al.3 Theoresolution in the 1961 spectra was about 100,000, or 0.03 A; this was five or The authors are with the E. O. Hulburt Center for Space Research, Naval Research Laboratory, Washington, D.C. 20390. Received 28 October 1966. more times less than the theoretical resolving power and was a limit set by aberrations in the lens system and by the slit width. The resolution attained with the echelle spectro— graph was more than an order of magnitude greater than other experimenters have obtained in this wave- length range; Kacalovo and Jakovleva4 covered the range 3100 A to 2470 A using a 1-m, 600 lines/mm grating spectrograph, flown in a rpcket in the USSR; they obtained a resolution of 0.3 A in certain regions. With the same instrument in a later flight, Kacalov et al.5 attained 0.15 A resolution from 3100 A to 2700 A using the second order. Behring et (1U, employing the grating spectrograph described by Miller et al.7, in which a 600 lines/ mm, 50—cm radius grating was illuminated at 495° incidence, photographed the spectrum from 3000 A to 1800 A with a resolution of approximately 0.5 A. The excellent solar atlas of McAllister8 was compiled from the latter spectra. II. The Echelle Spectrograph The completed echelle spectrograph is shown in Fig. 1; it is installed in a biaxial pointing control of the type produced by the Ball Brothers Research Corporation of Boulder, Colorado. In operation, the entire nose cone of the rocket is raised through a distance of more than three feet by an electromechanical drive, thus un- covering the spectrograph and the mechanism for point- ing it at the sun. Stabilization against roll of the rocket body is accomplished by rotating the entire nose section with an electrically driven servo system sensing the sun with phototubes. The rotation is made possible by a large ring bearing, located at the last break in the rocket body, that may be seen just above the top of the March 1967 / Vol. 6, No.3 / APPLIED OPTICS 365 Fig. 1. The echelle spectrograph, mounted in the biaxial pointing control of the Ball Brothers Research Corporation (BBRC), in the forward section of an Aerobee-150 rocket. The nose cone is lifted to expose the instrument, as in flight operation. M. W. Frank of BBRC, left, and J. D. Purcell, right, are about to test its operation in the New Mexico sunlight by spinning it on the motor—driven turntable and testing its ability to lock on the sun. turntable on which it is mounted for simulating roll of the rocket when testing the unit. The instrument itself is carried on a transverse bearing that allows it to be pointed at the sun regardless of the yaw and pitch of the rocket, within certain limits. This also is accomplished with a servo system—and solar sensors. After com- pletion of the exposures, the instrument is once again lined up with the rocket, and the nose cone is brought back into its original position to cover the instrument and to protect it during impact. Recovery is accom— plished by means of a parachute. The echelle spectrograph itself, With the cover re— moved, is shown in Fig. 2. Dimensions of the case are 13.3 cm X 15 cm X 80 cm. The cylindrical tube on the front end contains the predispersing system used to eliminate stray light. Figure 3 shows the predisperser removed from the tube. The optics are mounted on a short bar whose underside is machined to a curvature that just matches the inside of the cylindrical tube. This instrument was flown and recovered successfully in 1961 and flown again in 1964. On the latter occasion the parachute failed to operate, and the instrument was demolished. The film, however, was recovered. On the original flight in 1956 the instrument was not mounted in a biaxial pointing control but was installed rigidly in the nose section of an Aerobee—150 with Optic axis parallel to the longitudinal axis of the rocket; light 366 APPLIED OPTICS / Vol. 6, No. 3 / March 1967 from the sun was directed into the instrument by means of a servo operated coelostat mirror. The spectra photographed on this flight, although extending to about 2700 A and producing new data on the contours of the Mg II doublet near 2800 A, fell far short of what the instrument could have produced because the rocket failed to ascend all the way through the ozone layer. The optical diagram of the echelle spectrograph is shown in Fig. 4 in plan and elevation views. Basically, the spectrograph is a Littrow system With a prism crossed with the echelle grating in order to separate the different orders of diffraction produced by the echelle. Light from the sun enters the slit S, is collimated by lens L, passes through prism P, is reflected and dis— persed in one plane by the echelle, passes once again through the prism which introduces the cross disper- sion, and is brought to focus by lens L on the film, located at position F. The design of the lens presented a difficult problem, since it was required to coverothe wavelength range from 3000 A at least to 2000 A and produce a sharp image. It was necessary to minimize the longitudinal chromatic aberration, and also the zonal spherical and spherochromatism. The lens was a triplet of 400 mm focal length and 65 mm diam, designed and constructed by the Farrand Optical Company.9 The negative center component was of crystal quartz, the outer posi— Fig. 2. The echelle spectrograph with the cover removed. Sunlight enters from the right through the predisperser system inside the cylinder. The echelle, prism to produce the cross- dispersion, and Littrow lens are seen at the left. Film from the supply container at the upper right is carried to the right around a roller, across the curved plate that determines the focal surface, around another roller, then to the left and into a cassette. Fig. 3. The optics of the zero-dispersion double-monochromatot predisperser, removed from the cylindrical housing. At the extreme left end is the final lens that images the sun on the slir. Fig. 4. The optical system of the echelle spectrograph in plan and elevation views. Lens L is an achromat of quartz and LiF, prism P is of fluorite, and field flattener L' is of quartz. tive components of lithium fluoride; the quartz com— ponent was mated to one outer component with glycerin and was air-spaced from the other. In practice, the lens was found not to be sufficiently corrected for use at full aperture f/ 6, and in flight it was stopped down to f/ 10, for which the circle of confusion was about 30 It in diameter. The lens was constructed and meticu~ lously adjusted and tested by Nolan and Barcus. In order to make use of 35 mm roll film it was neces— sary that the focal surface be kept straight in one direction. In the other direction the film was made to conform to a curve that optimized the focus. In the direction of the echelle dispersion, there was so little wavelength span across each order that the changing chromatic aberration was not a problem, and the focal surface was almost flat. However, to make it exactly flat it was necessary to introduce a field corrector lens L’ of crystal quartz. A hole was cut in this lens to permit passages of the beam from the entrance slit to the main lens. In the direction of the prism dispersion there was considerable chromatic aberration because of the great range in wavelength. This was corrected by feeding the film over a plate cut to the proper curve, established empirically, to correct for the residual longitudinal chromatic aberration, which was quite severe, as shown by the curve of Fig. 5. The echelle gratings used in the instrument were ruled by Harrison at the Massachusetts Institute of Technology and were replicated by Bausch & Lomb. The ruling parameters were 73 grooves/ mm and step ratio two to one. An angle of incidence of 63.25° was required to place the blaze on the center of the optical system. The echelle itself, of dimensions 68 mm X 145 mm X 25 mm, was mounted in a cast alum- inum cell of the proper shape to hold it at the desired angle. It was held firmly in position, by screws located such that they did not cause distortion but were de- signed to prevent it from shifting as a result of accelera— tions experienced during launch. Various spectra obtained with the instrument are shown in Figs. 7—11. The many orders produced by the echelle were separated by the 30° calcium fluoride prism. The linear dispersion of the system, propor— tional of course to the order pumber, yaried from 2.5 A/ mm for orgler 60 near 4000 A to 1.25 A/ mm for order 120 at 2000 A. Orders overlapped from end to end by an amount that increased the shorter the wavelength; for example, betweeno orders 80 and 81 the overlap was approximately 04 A, while orders 105 and 106 over— lapped by about 8 A. The overlap was of considerable assistance in establishing the wavelength scale. In the cross direction, using a slit length of 0.25 mm, orders were just separated at 4000 A and were separated by more than the slit length at the short wavelength end, owing to the increasing cross dispersion produced by the prism with decreasing wavelength. Focusing and adjustment of the instrument were tedious, and the empirical determination of the proper curve for the film guide was especially so. Final focus- ing was always carried out in vacuum, since the instru— ment was evacuated in flight. When making adjust- ments in air, allowance had to be made for the change in refraction produced by air at the various interfaces; this caused a marked change in focus and a wavelength displacement of nearly 1 A. In the 1961 flight, Eastman Kodak Type IV—O, UV l4 A o 000 :2 8 3 32° 5 B 8 '0 mm: ,c ~IO~ N V l ' 'V o v (I) >7 8 <8~ m g V 9 /' «6w :9 g o . x 8 g l < Z _. .J <4 N <1 I— v 2 Q I— O 2-— % l l I I l l l l I l 2 4 6 8 IO 12 I4 I6 I8 20 22 DISTANCE TRANSVERSE TO AXIS (mm) Fig. 5. The shape of the focal surface required to correct for the longitudinal chromatic aberration after two passes through the lens. March 1967 / Vol.6, No. 3 / APPLIED OPTICS 367 FEFIV PRISMS PLANE MIRROR / K2555: LENS PLANE MIRROR SPECTRUM Fig. 6. The optical system of the zero-dispersion double—mono— chromator employed as a predisperser. Stray light was rejected by introducing the knife edge into the intermediate focal position in order to intercept all radiation of wavelengths longer than the desired limiting value. ‘—-l0A—- IIIIIIIIIII ‘I - I>I"|- <3: I,' -3200 ORDER NUMBER I00- I05- -2300 I I I I I I I I I I I ~——Io ——. Fig. 7. The spectrum of a low—pressure rf—excited iron arc, as photographed by the echelle, flown in 1961. sensitive film 35 mm wide was used, since extreme resolving power was the most important consideration. With a slit width of 15 y, there was sufficient intensity in the solar spectrum to obtain ample exposure with this emulsion during a rocket flight. It was necessary to coat the film base with a rim—jet—black backing to avoid static-electrical fogging when transporting the film in vacuum. For the 1964 flight, Type III—O UV film was used, because IV—O, regrettably, was no longer avail- able. This resulted in somewhat more noise associated with granularity, and the slightly greater speed was not needed. Twelve exposures were made on each flight by means of a programmer, driven by an electric motor. They ranged from 2 sec to 84 sec in 1961 and 2 sec to 96 sec in 1964. When transporting the film, a shutter blocked the slit to avoid any possibility of fogging. The most difficult problem associated with an echelle spectrograph, particularly one of the Littrow type, is the stray light. Since the intensity of the solar spectrum is of the order of 100 times greater in the visible than at 2000 A, the short wavelength limit which it was hoped to reach, and since the depths and profiles of the strong Fraunhofer lines were of particular interest, it was necessary to reject the unwanted part of the solar spectrum before it entered the slit. The first method was the low pass optical filter devised by 13mm- berg“), and by Regener”, consisting of a pair of quartz prisms separated along a diagonal with a liquid having a dispersion relative to quartz such that wavelengths longer than the desired cutoff were totally reflected out of the prism, while wavelengths shorter than the cutoff 368 APPLIED OPTICS / Vol. 6, No. 3 / March 1967 were transmitted with only a small reflection loss. Although this prism was found satisfactory by Regener for removing stray light from a stratosphere balloon- borne spectrogrgph designed to record the solar spec- trum to 2500 A and below, it did not prove to be sufficiently effective for the echelle spectrograph. Furthermore the cutoff wavelength varied rapidly with temperature, resulting in a diflicult problem of control. The prism recently designed by Crosswhite12 oflers promise of satisfactory low pass characteristics, but in the form tested, where only four internal reflections per prism were used, the stray light transmittance was not reduced sufficiently. The actual device employed for eliminating stray light was a small zero—dispersion double—monochroma— tor of quartz, located in front of the slit, with an optical system designed by Purcell (Fig. 6). The entire device fitted into the cylindrical tube shown in Fig. 2, 4.45 cm in diameter and 20 cm long. Two Féry prisms were used. The first prism dispersed the beam and focused it into a spectrum of solar images in the inter- mediate position. The second Féry prism recombined the spectrum and collimated the beam. With the knife edge removed, a white light image of the sun was produced on the slit by lens L. The two plane mirrors simply folded the system into the compact form re— quired for use in the rocket. The intermediate lens served to image the first Féry prism on the second. When the knife edge was introduced in the intermediate focal plane, the portion of the spectrum to the red of its position was cut out; the reconstituted solar image then contained all the wavelengths to the blue of the limit determined by the knife edge. Also mounted on the same base plate, but not shown in the figures, was a simple boresighting device. This comprised a small lens, a plane mirror reflecting the converging beam of sunlight back toward the sun and focusing a solar im- AIIIIIIIIIIIIIIIIII LOG INTENSITY I I I I I I L ,I I I I r . * I I I - I00 0 I00 200 Ai— mIlIi ANGSTROMS Fig. 8. The profile of Fe II 2739.545 A recorded with the echelle flown in 1961; for comparison there is also shown the profile of the 6.65-m concave grating instrument used by Briickner to produce the Gcttingen Solar Atlas, measured with Kr I 4274.97 A, and scaled to 2700 A. Solid curve: echelle Fe II (63) )x 2739. 545 A. Dotted curve: Gottingen Kr 4273.97 A, scaled to 2740 A. 1o I111111111l ORDER 105~ _ , (“Wei-films“ ~2300 ' _ a . ¥’” ' 1‘ , * ’ ME) 110- ‘ .. - A... ...___..., , 72200 I I I I I I l l I I I 10A )\__.. Fig. 9. The spectrum of a carbon arc crater, photographed with the echelle spectrograph flown in 1961. Ghosts and stray light associated with the 2478 A line of C I are clearly seen, though not of great intensity. At long and short wavelengths the bands of CN from the arc stream are prominent. age on a ground-glass screen and reticle, located on the front where it could be easily viewed. This was aligned with the predisperser and was used to align the entire instrument with the solar sensors that controlled the pointing during flight. When flown in 1961, the knife edge was set to block off the spectrum longward of 3300 A. Although it might appear that the cutoff was gradual, because the intermediate spectrum consisted of images of the entire sun, this was not the case because the spectrograph entrance slit and the knife edge position for one par- ticular wavelength were at conjugate foci. In spite of the double dispersion, there was sufficient light leakage through the predisperser to secure a well-exposed spec— trum all the way to 4000 A. General, residual stray light limited the range that was covered; at the short wavelength end below order number 110 at 2200 A almost nothing could be detected above the background fog level. In the 1964 experiment an echelle with im- proved stray light characteristics was available, and the predisperser knife edge was set to eliminate the spec— trum longward of 2500 A. As a result the spectrum was extended to 2090 A before it became swamped by stray light. To go beyond this limit is especially difficult because of the sudden decrease in intensity of the solar spectrum by a factor of about five, from 2100 A to 2090 A. The echelle grating was re—aluminized before flight to maximize its reflectance. In order to accomplish this properly it was necessary to evaporate the alum- inum in a direction as near as possible to the normal to the echelle step—surfaces, the condition which produces maximum reflectance of the coating. But normal incidence on the steps would have required 63° incidence on the echelle, and the result would have been a change in the thickness of the coating from one end to the other because of the difference in distance to the aluminization source. This would probably have introduced phase changes and loss of resolving power. The compromise arrived at was 35° to the echelle normal and about 23° to the step—surface normal. Although the absolute transmittance of the instru- ment was not measured, the speed was high at A > 2300 A. At shorter wavelengths the transmittance decreased rapidly because it was not possible to obtain lithium fluoride that was completely free from absorp- tion near 2000 A at the time when the lens was con— structed. The over—all transmittance of the system relative to 3000 A was estimated at 5% and 1% at 2100 A and 2000 A, respectively, not including the grating efficiency. The performance of the spectrograph in the labora- tory can be judged by reference to Fig. 7, the spectrum produced by an rf—excited low—pressure iron are of the type described by Corliss et al.13,having narrow lines free from pressure broadening The resolution was determined to be 0.030 A by studying the profiles of several of these 1ron lines. For example, in Fig 8, the profile of Fe II 2739. 545 A 1s reproduced. Introduced for comparison is) the profile of Kr 4274. 97 A, scaled linearly to 2740 A, as recorded by the 6. 65 m concave grating spectrograph employed by Briicknerl4 to pro- duce the Gettingen Atlas of the Solar Spectrum. The width at half-maximum of the instrumental profiles of both spectrographs was 0.030 A. The principal differ— ence is that the echelle did not produce a really clean spectrum. The profiles are alike down to about 10% of peak intensity; below this level the line shape of the echelle spectrograph soon becomes broadened and the stray light level never falls below about 1% of the peak intensity. The stray light characteristics and speed anomalies are best shown by Fig. 9, the spectrum of a carbon arc crater in air. The carbon arc radiation is quite similar to that of the sun in distribution, following a graybody WAVE L ENGTH (ANGSTROMS) 1 1 1 1 1 [ 1 1 1 l «10 ANGSTROM$—> X—" Fig. 10. The solar spectrum, photographed with the echelle spectrograph from an Aerobee—150 rocket on 29 August 1961. The reproduction is a composite of prints of several exposures. March 1967 / Vol. 6, No. 3 / APPLIED OPTICS 369 ‘———-101‘i———~ I1111l111|l ORDER NUMBER 0 A (A) lilllrulrrl IOA Fig. 11. The short wavelength end of the solar spectrum, photographed with the echelle spectrograph on 21 November 1964. This is the extreme limit reached with a resolution of the order of 0.05 A, and marks the end of the region where the solar spectrum originates from the photosphere and is characterized by Fraunhofer lines. curve of 3800°K through this region. Conspicuous In two orders 1s the carbon line at 2478 A. Below this wavelength the spectrum gradually becomes compli- cated and is largely produced by band emission from cyanogen in the arc stream. The long wavelength cyanogen bands are also seen, in spite of the predis- perser cutofi at 3200 A, which shows clearly at order number 75. Since long wavelength stray light was largely removed by the predisperser, the residual stray light, except at the short wavelength end of the spectrum, was so-called nearby straylight, which consists of ghosts of various sorts. They can be seen best by examining the series of images of the 2478-A line. Since the ghosts have the same wavelength as the parent line, they are deviated when passing through the prism by exactly the same amount as the line. Therefore, they lie along a straight line that is slightly inclined to the direction of the echelle dispersion, since each echelle order is deviated increasingly with decreasing wavelength from right to left as a result of the increasing cross dispersion of the prism. In addition to a number of clearly defined ghosts, there is nearby stray light consisting of many less distinct ghosts, some close and some far from the parent line, together with general “grass” which extends all the way across between the images in different orders. As a result, it is extremely difficult to make an exact correction for stray light by examining the density be— tween orders, since the stray light is associated in a complicated way with the intensity in the spectrum. Generally, the nearby stray light increases rapidly in intensity with decreasing wavelength, since it follows the rule for ghosts—proportionality with the square of the order. There is no cure for this defect other than to produce an echelle which is more perfect. In the echelle spectrograph of Detwiler and Purcell15 working in order number 201 at the Lyman-alpha line of hydro- gen, 1216 A, stray light from this source was so strong that the efficiency of the echelle was extremely low be- cause most of the light went into the spectrum alongside the line and between different orders of the line rather 370 APPLIED OPTICS / Vol. 6, No. 3 / March 1967 than into the line itself; at Lyman-alpha a rough esti- mate indicated that as much as 90% of the line was lost between orders. The transmittance of the optical system was quite nonuniform along each order, from center to edge, as can be seen in Fig. 9. There were two reasons; first the echelle was adjusted so that the blaze came at the center of each order, and the blaze was rather sharp in character; second, at the shorter wavelengths where the transmittance of the optical materials was low, the in— creased path for the skew rays through the optical system that reached the edges of the field resulted in a further reduction of transmittance. The solar spectrum photographed on 29 August 1961 is shown in Fig. 10, a print already published in many journals and on the cover of APPLIED OPTICS for November 1962. This reproduction is a composite of prints of different exposures since no single exposure could cover the entire wavelength range because of the great decrease in intensity of the solar spectrum to short wavelengths. The spectrum obtained on 12 November 1964, al- though having slightly less resolution than the 1961 spectrum, reached shorter wavelengths before merging into the stray light. A section of one of the 60—sec exposures is reproduced in Fig. 11. Examination of Figs. 10 and 11 shows how the gen- eral, nonfocused stray light, coupled with the decreased speed of the instrument and reduced intensity of the solar spectrum, determined the short wavelength limit that could be reached. Still shorter wavelengths could have been recorded if optical materials of greater trans- parency at short wavelengths had been available, and if the echelle efficiency could have been increased by better aluminization and further reduction of stray light. Adjustment of the predisperser to reject even more of the spectrum would also have reduced the back- ground still further, but the instrument was demolished before these changes could be made. To illustrate the quality of the 1961 solar spectra, a C11 (63) F! I (l8l) F: U (4|?) -4 LI] ? Fl I (95) .> Cr 1 (63) F. I (46) Cr II (6) F. I (51 ——. v n (13) ——Mn 1 ... Fe H46) Fe II (63) Cr II l6) % <4 / S III (l6) ‘/ 1.1 *3 Fe II (260) ”,7 4 Fe 11(62) L. NW < l 2739 2740 2741 2142 2743 WAVELENGYH (ANGSTROMS) Fig. 12. A short section at the center of order number 89 of the 1961 echelle, greatly magnified, together with microphotometer traces of two exposures. Table I. Lines in Echelle Spectrum 3000—2091 A Lines Lines Percent Intensity listed identified identified 1 1943 457 24 2 1516 679 45 3 1275 783 61 4 1 104 836 76 5 7 12 614 86 6 393 367 93 7 139 132 95 8 50 47 94 9 14 14 100 Total 7146 3929 55 short section near the center of order number 89 is re- produced in Fig. 12, greatly magnified. Included also are microphotometer traces of two different exposures. The finest structure is obvious granularity. Most of the identifications are positive, but the coincidence of a line with Li I, 2741. 204 A, is considered to be accidental, since Li has an extremely low abundance in the sun. The identification of SIII is in some doubt, because Fraunhofer lines of no other doubly ionized ions have been recorded. However, most of the lines of S III that might be expected were present. The many extremely intense, broad lines of FeI and Fe II, such as Fe II, 2739. 545 A, form the most characteristic features of the solar spectrum in the middle uv. The reduction of these spectra to obtain precise wavelengths was performed empirically since the changing magnification in the lens system with wave- length and the distortion across the field made the pre cise application of the grating equation impossible. It was a simple matter, however, to identify many ex- cellent lines of Fe I and Fe II for use as standards. This was done for each order. The difference between air and vacuum wavelengths had to be taken into account, of course, since the actual solar exposures were made in vacuum. After selection of the standards, a cubic equation was used to fit the measured line positions and the dispersion curve. Overlap regions between ad- jacent orders were carefully checked. In the case of sharp, well—defined lines, wavelengths were determined to an accuracy of 0. 01 A. For faint or diffuse lines the error was approximately 0.020 A to 0.025 A One diffi- culty in making the measurements with a microphoto- meter or comparator arose from the fact that each order not only slants slightly relative to the normal to the slit, but also is slightly curved because of the action of the prism and the field corrector. Weak Rowland ghosts, though present, generally fell between orders, as discussed earlier, and did not interfere. The analysis of the solar spectra to determine wave- lengths has been completed, and as many of the lines as possible have been identified. Table I lists the number of lines recorded, grouped by intensities on a visual scale of 1 to 9, corresponding roughly to the scale —3 to 25 used in the recent revision of the Rowland Tables prepared by Moore et al. 16 Comparison with the latter line list over the region 3028 A to 3069 A resulted in excellent agreement in wavelength, intensity, and identification. The average wavelength deviation for the best lines was 0 003 A, and 0.014 A for poor lines. The faintest lines of intensity 1, which are close to the noise level of the spectra, were found to be the least certain, of course. Thirty—four of the sixty-four faint lines in this wavelength range were coincident with lines in the list of Moore et al., whereas fourteen of their fifty-seven weak lines were coincident with ours. In a similar comparison with traces made by Migeotte17 on the J ungfraujoch, fifty—two of the same sixty—four weak echelle lines were coincident, while forty—four of the fifty—seven lines appearing in Migeotte’s spectrum were coincident. These differences are, in part, attributable to the fact that the background continuum, even on the shortest echelle exposure, was greatly over-exposed; therefore, the weak lines in the solar spectrum were difficult to detect. It is expected that the 11v solar line list, derived from the 1961 and 1964 echelle spectra, will be published by the Government Printing Office in 1967. It will be of the same general form as the Preliminary Rowland Tables of St. John et al.18 and the new revision by Moore et al.19 With the aid of the Ultraviolet Multiplet Ta— bles by Moore, and other material, about half of the observed lines were identified—but no more. This shows that a need exists for much additional research in laboratory high resolution spectroscopy, especially for the abundant solar elements. Until a much larger fraction of the lines are identified, it will be hazaldous to place confidence in the identifica— tion of some particularly interesting lines, for example, Hg I, 2536 519 A. Mercury has not been detected in the sun; there is a line at 2536. 50 A in the echelle spectrum which is faint and within one standard deviation of the mercury line; but in view of the large probability of a chance coincidence, this cannot be identified as the mercury line. The reduction of the echelle spectrum to intensities .. massaging Fig. 13. A portion of order number 73 of the 1961 echelle spec— trum, centered three-fourths of the way to the right and en- larged about seven and one-half times. The heavy curve is a linear plot of intensity, as reduced from the echelle spectrum, and the light curve is the solar intensity from the G6ttingen Solar Atlas, recorded with a 6.4-m concave grating. Upper curve: Géttingen Atlas. Lower curve: NRL rocket. March 1967 / Vol. 6, No. 3 / APPLIED OPTICS 371 is difficult because of the problem of correcting for the stray light background. An attempt was made over a short region, 3330 A to 3338 A, which 1s covered both by the Géttingen Atlas and the echelle spectrum This is reproduced in Fig. 13. Line intensities range from some of those listed as —3 on the Rowland scale, which are among the faintest listed by St. John et al.13, to lines of intensity 8. The echelle film was calibrated by con- structing an H and D curve from exposures made with a carbon are. The difficulty in application of this curve, however, was making the proper correction for stray light. By employing the background density as a vari— able parameter, it was possible to obtain a nearly perfect match with the G6ttingen trace, as can be seen from Fig. 13. However, it was not possible to relate the stray light correction so determined to any obvious posi- tion in the echelle spectrum. Therefore, there was no straightforward way to obtain the stray light correction below the limit reached by the Gfittingen Atlas (2965 A) by examination of the spectra. The procedure being followed to reduce the echelle spectrum to intensities from 3000 A to 2100 A, where high resolution spectra that are free from stray light are not available, is to compare the echelle spectrum with the low resolution spectra of Wilson et (11”, and of Malitson et (11.21 These spectra have some twenty times less resolution, but they are fairly free from stray light. The echelle spectrum is recorded with a micro- photometer in digital form on magnetic tape so that the densities can be transformed with a computer. The microphotometer trace will then be subjected to an averaging process in order to reduce the resolution to the value of the earlier spectra; then the stray light correc- tion parameter in the H and D curve will be adjusted so that the average echelle spectrum matches precisely the earlier spectra. After this is done, the detailed microphotometer trace of the echelle spectrum will be reduced to intensities using the adjusted H and D curves. It would be wishful thinking, of course, to hope that the echelle spectrum could ever be corrected to show the cores of the deepest Fraunhofer lines: once they are filled in by stray light, this information is lost. To ob- tain high dispersion spectra with no stray light at all it will be necessary to employ a large grating spectrograph of perhaps 2 m radius of curvature, equipped with an ex— cellent predispersing system. An instrument of this type is being constructed for flight in the Astronomical Telescope Mount Mission of the Apollo Manned Space Flight Program. It will be mounted in such a way that it can be pointed with precision at any position on the sun or in the chromosphere. One of the astronauts will make the exposures and bring back the film. Many individuals have made significant contribu— tions to this program. To the Farrand Optical Com- pany we are indebted for the expert design and fabrica- tion of the lenses and prism. Aircraft Armaments, Incorporated, constructed the coelostat used successfully for the 1956 flight, and the Ball Brothers Research 372 APPLIED OPTICS / Vol. 6, No. 3 / March 1967 Corporation produced the biaxial pointing control employed in 1961 and 1964. G. A. Vanasse and later D. J. Michels and A. Boggess III were involved in setting up the instrument. The special, technique for aluminization of the echelles was developed by D. W. Angel. Reduction of the spectra was commenced by H. P. Erickson and carried to completion by L. Giddings and P. A. Simmons. For the echelle gratings we are particularly indebted to the late David Richardson of Bausch & Lomb, Incorporated. References 1. 2. {/3 G. R. Harrison, J. Opt. Soc. Am. 39, 522 (1949). J. D. Purcell, A. Boggess III, and R. Tousey, I.G.Y. Rocket Report Series, No. 1, p. 198, Natl. Acad. Sci-Natl. Res. Council, Washington, D.C. (1958). J. D. Purcell, D. L. Garrett, and R. Tousey, in Space Research III, W. Priester, Ed. (North-Holland Publishing 00., Amsterdam, 1963), pp. 781—786. 4. V. P. Kaéalov and A. V. Jakovleva, Izvestia Krymskyi Astrofiz. Obs., Akad. Nauk, SSSR 27, 5 (1962). 5. V. P. Kacalov, N. A. Pavlenko, and A. V. Jakovleva, Izvestia Akad. Nauk, SSSR, Ser. Geofiz. No. 9, 1099 (1958); Bull. Acad. Sci., USSR, Geophys. Ser. No. 9, 636, Izvestia Akad. Nauk, SSSR, Geophys. Ser. No. 8, 1177 (1959); Bull. Acad. Sci. USSR, Geophys. Ser. No. 8, 844 (1959). W. E. Behring, H. C. McAllister, and W. A. Rense, Astro- phys. J. 127, 676 (1958). S. C. Miller, Jr., R. Mercure, and W. A. Rense, Astrophys. J. 124, 580 (1956). 8. H. C. McAllister, “A Preliminary Photometric Atlas of the Solar Ultraviolet Spectrum from 1800 to 2965 Angstroms”, Univ. of Colorado Printing Services, Boulder, Colorado (1960). 9. S. Rosin, J. Opt. Soc. Am. 42, 451 (1952). 10. 11. 12. 13. 14. 16. 17. 18. 19. 20. 21. E. M. Brumberg, Compt. Rend. URSS 2, 467 (1935). V. H. Regener, Quart. J. Roy. Meteorol. 800., Suppl. to Vol. 62, 9 (1936). H. M. Crosswhite, private communication. C. H. Corliss, W. R. Bozman, and F. O. Westfall, J. Opt. Soc. Am. 43, 398 (1953). G. Bruckner, Photometrischer Atlas des Nahen Ultravioletten Sonnenspektrums, 2988 141—3629 .4 (Vandenhoeck and Ruprecht, Gottingen, 1960). C. R. Detwiler and J. D. Purcell, J. Opt. Soc. Am. 52, 597 (1962). Charlotte E. Moore, M. G. J. Minnaert, J. Houtgast, “The Solar Spectrum 2935 A—8770 A”, NBS Monograph No. 61 (1966), U. S. Govt. Printing Oflice, Washington, D. C. M. Migeotte, private communication. C. E. St. John, C. E. Moore, L. M. Ware, E. F. Adams, and H. D. Babcock, “Revision of Rowland’s Preliminary Table of Solar Spectrum Wavelengths with an Extension to the Present Limit of the Infra-red”, Carnegie Institution, Washington, D.C. (1928). C. E. Moore, “An Ultraviolet Multiplet Table”, Natl. Bur. Std. Circ. No. 488, US. Govt. Printing Office, Washing— ton, D.C. (1950). N. L. Wilson, R. Tousey, J. D. Purcell, F. S. Johnson, and C. E. Moore, Astrophys. J. 119, 590 (1954). H. H. Malitson, J. D. Purcell, R. Tousey, and C. E. Moore, Astrophys. J. 132. 746 (1960). 2239 ”mull: .lnllllnll'l III" I III-l. All 2235 lllullllllll 223! 2227‘ JIIJXLJX 2239 225! 2247 2243 10 U ' 'fl I ' l I ‘ Vl‘l I II I’ I I | - ' II ' ll 1 I' I: I ' I If '| I I , I; I ||: | l l l l 1/ fl . | V 1 l \ l J l I H ‘ l \f I i L, j i I, \ II ‘ l Iv‘ V ‘ l l I I‘ {\I \ A ‘\ I ‘IA I I I I I v I I, V f L II I I ' V 1"] I a llllllllllllllLIllI IlllllLll lilllilll lilllljlllll‘llllli Ill IlllLJlllllllll llllllllllll llllll |||111L4Ll||||lllIllllllll] 2250 2254 2258 2262 11 ~.__ lLlllllllllllllllll lllllllll 2264 lllllllllllllllllll llllllLllLllllllllJlLllllJllllllllllllL $5M \‘ k: ,Jl J u \\ I J .«v I U H fk’ I A llllltlll lllllllll lllllllllll Jllllllj 2268 2272 12 l I l l 1“ IA \ l l I \ \ I IX! ’4 fl 1‘ \ : I l I H II I l \ r I : I II I \, l I II / ‘I ‘ I II I ‘ l t | I l I \\ I l. 1 l . f n I f i l” 1 l I / ‘l I ' , A / v N f I " ‘ I " \I \ t ‘b \l x ‘l O I l l I l l llllllll l I I I l 1 I l l | J l l l I l l l l I l I I l lllllllllllllllll I I l llllllllllll I l 1 l I l l l l I l I l l I l l lllll I I l 2270 2274 2278 2282 13 All I“ Il‘l lllll 228! 2293 2289 2285 14 p.‘Il‘l I’ll 2287 l 1 229! 2299 2295 15 |‘I||I lllll'l'|l||'ll|l llllLJllllll llllll 2304 232 2308 2300 16 2.0 17 -------------------------t 'l'l. ' l.’ ’Ill llilll h'-’ I. 'I‘, 2336 2328 2324 ZI) L5 2332 18 0.5 I, Ill]! 1’ “I m I, I)! 11:1HHHHHHHHv A”! [I In! 2359 ------------------------d 2355 255! 2347 L5 L0 |.5 L0 0.5 M * J3 HM \M M M Ma : 3 n : .' : .5 MK fl 1 i W b WM "a” ‘K: M, L M v ......... || I I Jll 0.5 -..----------------..---J Illlllnlll lllll ‘\| I. I, I I Illlu‘llt ‘u‘ I.|I.Il.| I'll ll.l L5 L0 238l 2377 2373 2369 23 2.0 LO 26 2.0 0.5 \- 27 llllll IIIII|I| I'll llllullllllll'lllllll'lll Illllllullhlllv IIIIIII lllll Alllulll IIII‘I‘I llll' |lgl||‘.l' L5 L0 I I I 1 ‘. I J I' I‘. F '1 I] h. I ’ N A H ‘ \‘ V‘ 'I l "I . “II II {1.1 “J‘ II I I W ' I' ‘M‘ ‘ J” " I . i ‘H: \n ‘ I I .' \ I l ‘\ \/ i4 | “1' \ [I | I I l A ,\ .4 ”v1 I' ‘I / II ‘ '\ I' x ,' ' I ,I ' Hll I I"! r v \ H I I II I J I i I I I l l | 1' || \‘ I! I, ‘ I J ‘ J l II VI :. \ If ‘ ‘ V I ' ‘ I I I I: ' I \ l 1 I \\ ‘ I I I g I I I l I I I I l I ‘ I CL5 V U lllllllll‘llllllillIJJJIIIII lllllllll lllllllllllllllllllIllllJJll llllllllll|lljlllllllllllljlllllllllillllllLlllllllJ_Jllllllllllll 2448 2452 2456 2460 33 '5 0.5 34 |.O 0.5 Illllll'll l'I" ll, - [III II III" IIIII‘III I'll ll'lill Ill ['1 I'll Illlal ll \‘ I‘ I|II| Illl I‘“ I‘ll 2485 24m 2477 2473 L5 015 36 Ill I, I I‘ llllll llllln||u|.|\ \|| \\ 'I III! I 2494 2490 2486 2482 L5 (15 37 ||||||||||| .- I'll!- I]!!! I IIIIIIIII 2504 2500 2496 2492 L5 L0 38 lll'l I l'lll | lll'nrllllllflulllflllhlllllllult IIIIIIIIWIIL ll") ,l’Illlll IIIIIIIIIIIIIIHHU .IlanIIIIIIII IV 1l||lllllu. sllhfltllll ll. llHHnllhi nl IIII 250 2506 2502 V .lv‘llnlv‘ 2498 |.\I.Iu .l.al|||\ ‘ullli‘llll fill! I IIIIIIIIiV \Ilullul'Jllllllllllllullullxl‘ll IIIIII |||I|IIIIHIHIHIHV k. IIIIIIIII 259 255 25” 2507 L5 L0 4O ------------------------d hl I‘ll l||.| IIIU IIII |.|.I|.\ “HM-“II: IIIII lllllllIIll]llllllIllljllllllllillLLillllllllllllllllllllI llllllllllllLlll llllllllJlll lllllllll llllllllllllllllll lllllllllllllll 25KB Jlll L5 L0 (15 O 2528 2524 2520 41 2.0 : I l I | N . I I A M A I f I n. Ilfl! n . n J I A l h I4 I : l 1 b M . I II: I ' I '1 III ‘I iii“, : A I’ ,‘ III\ II |I I I,"I I {III I II {I I ,’I 'I' ‘I (”1' I \. H I.' 'I Ii . I I II lI \J ' 1 b I I' ‘I' I {A} H i} . I .’ I I II \J V ......... 1W“... l II] [J 'll'l'ul ll llllllllll l I! I! 4||o| llllllllll ['llllllll'llll p-----.-.-.-------------- Illllll 2563 2559 2555 255! 45 "I‘I \Iv ‘\ 46 2572 2568 2564 2560 lullll'IU 49 7%: ”r “21> : ‘23> $1.2. / (I: {g} C j -- «:::. I; 4"” __-- C 5.0 2.5 WNW/WWW OIlllllllllllllIllllllllllllllLllllll|llll lJLJlllllllllll 111]] | I I l\ : l I I I l | I I I I L 2606 26|O 51 "'| '1' I lululllllllllllh. 2628 2624 2620 26KB 5() 52 5.0 3/ i E g g 53 P------------------------ .II "I “ ||||| 2646 2642 2638 2634 IO 54 ll'l‘l’ [\l' “In! U U I. ill-lull] F 2667 2663 2659 2655 IO 56 5.0 ’1 ------------------------‘ ”I ll." U "Irv Fill J fillll‘ulj ''''' M' A} All I III II Iv A}! I\I\IIII|||I‘.I|I|||.|II|IIII- “| l I l 'l\‘ J 1“ N " III I I'll, 'Illllllllnlalalalll\‘v III‘IIII' 2697 2693 2689 2685 IO 59 P------------------------ llflllllll lllllllllllw'llll' “II” IIIIIII III .I lIIIIIul'I H. ”V HHHIINNIII. Infill III. 'II IO 60 2704 2700 2696 2692 ,‘allllr II I. II '1' "ul‘ Ioal'.l|r|llllllnl|| II ‘l'u'lll,lil|lllu 'll'c l‘. III-III-..‘ ll'lllllllil I, ul||||| I’I'III ll‘lltll‘. ”IIIIIIIIIII ll‘l‘l‘l'l ; , “1|.th” I Illlll II.’ fill-lull" .‘ulll' Ill": lrl||lzr||ill I‘ II I...‘ All... “I.“ .lllllllf Ill. 1 IIIIIIIII'IIIIIIIHIIU nll lil IIz||\ I’llllu' |.||.|I|d “II .II‘IIIHJI '01‘1ll‘lrlll’lulllllllll illlll lll.‘|:" I]. U ‘l .|I|III.| lllnllolllllnl' :lfilllullllll' I. I. I I. \I‘IU Inl|||l|||| P U.‘ IO 61 275 27” 2707 2703 O llllllllllllllllllllliIIIIIIIlllllllllIllll|lI‘lllllllllIllllllIlllllllIllillllllJIIlllllllI_JJJJl|llIlllllllllllllllllllllllllllll‘lllllllll I‘I‘x‘ 62 2725 272| llll'lul. lIIMIIHHFIl 27|7 «HUM “In: I\|‘I|U RIII IL: QHHIIIIiHHHIIIIIII .l ll'll'l-II‘IIIIJ \llhl... IIIIIIIIIIIII | 1““.an ‘III '1' III 'lllll'llllllllllllll'l IIIIIIHHHH' IIJJIIII I .Illlllllll'l‘ll fiddlhlll'l I [I‘ll-ll. .ll [I'll]- IO 27l3 .I..||.|.V I'llll 'll‘l'llll.‘ III I{ IIIIIIII. I”. ”HHU “HHHI‘ I I'll! 273 2727 2723 2735 IO 63 II.I-I|II All\.l..l. 2746 2742 2738 2734 IO 64 -----....-.----.----.-.---+ \ ‘J llllljlllllllllllll _.——— -— ‘— Illllllll Illlllill u" o .. ...—— llllIllllllJlllllll llllllllllllllllllllllllllllllllllllllj ) lllllllll 1D __.. _——- IIlJJllll lllllllllllllllllll 2744 2748 2754 2756 65 IIII“II\II..| l‘ fil‘l‘ \ lllJIllIllllilllllllllllllllIIJIIIJIJJIIIIIIIIJILIIllllllll llll I'll!- “I‘ll fl"I'II.||.|I lllllllllllJllllllllLlllllLllJllllllIlllLLlllllllllllAlllllllJllllllllllJllllllII IO 5 O 2766 2762 2758 2754 66 l\ r: 'll’lll I'lll'." In II" Ill-Ill '- “III! II.“ fil ' I I. I] " 4'IIIU lll-Illl'l'. ll‘llll.‘ ’"HII IH l‘IIIl“ \ \ "l' I" IlillllllllllllllllllllllllllllllJlllllllllllllll IJIlllllllllllllllllllllllIllllllllllll 2768 lll|ll JllllllllJlJIllllllllll IO 0 2776 2772 2764 67 I I I 1 I I l l I l l | l 2777 I" I A. I III'. I ll Inn II ,‘ I I | II I I J n' ‘J I I H I I \Vh — I I | \ l l ' I | I ' IIIIII IIILLLLIJ I I I I I I I l I IlIlIllJIllIIlllIlJ 278l IIIlLllll‘lIIIIIIIl ll4JllllI|IlllJJIll 2785 [A Vf‘lv‘ \A‘QM / HAN-u IlllIllJL IIIIJJIIIIIIIIIIIII 2789 68 5&) 2.5 2786 2790 [M , I .I. 'l I I I I I I I' I Ni I I ' A: I! I I I I I I |\'l [‘1 W‘f I I IIIII'I‘l/'WIIIIJ\I\ . I I In ' I A4 9 {H II"! I” II’I,’ u '..' “I 1‘ V 'I' I . U WM; fW:’lA\\{M v ,".{ \ 'v OllllllIllllllllllllllllllllll lllllllll |llllll|lllllllllll lllllllll lllllllll llllIIIlIlJJJllllllllllIJllll llllIlJlI llllllLllJlllllllll 2794 2798 69 5.0 / ‘q- .. ____________—_.—~/" C\ l l l I [I1] I fli|l l H : ' 4 l l a \ .l I Nb 3 I' * 1' Ill \ I, i \ u A' l I “ IL' '1” 'l H U l~ |l \ A /’ ‘1 , J . h "\1 “WW I l , LIV 11“; fiv'\r ‘ "I \V \ I i‘, I 11|1l|.| Ill lllllllllllll lllLllIlJllllllllllllllilllllllllllllllIliillllllllllLlllllllllllllllllllllllllllllllllllJlllJi 2800 2804 2808 70 IO v” 1‘ l \ ~ : I f \//f V "KJN IfV-\ W\~/\l I" I ‘l I, ‘ V ‘ LvW~ M W II v V ‘\ I?" \ ~‘l ‘J ‘ .« W- III||l|llIll]lllllllllllllllllllllIIIllIIlllllllllllllllllllIlllllllIlllllllllll|IllllllllllllljllllllllllIlllIllllllllllllllllllllllllllll 2795 2799 2803 2807 71 flil’l .lv ||.| II I. AH” III. (IIIIIIIII‘IIIIIII H [IV [V a Ah] IINII‘III lulu Ill IlllllHnrluv IIIIlIlI'III. I III] r'l'lllll' III 'ltl‘llllll' ||||||||||||| lllll fillillllll \l IIIIII ll! ‘I'Il\ lllllll I III] I ll. lHHv \‘Illlll'lll |:\ll:..ll lllllllllllll “HWIIIIIIIIIIII II IIIIIIII'IIIHIII |.|.||l.| III lullllllllslil. Illllllllllllllll llllllllm Will. um, J 2820 286 2&2 2808 50 25 72 IIIIIIII Il-llul‘lll' I '1'” IIIIIIlII||l|||IIlL "Wu.” H v I! 5"! “n 1 5n 2830 2826 2822 288 10 73 fl llllllllll'llllu |||.|l. I II..|. Al ll/l llllllll .lI/II I. null||nl|ul'llllllllllalull|lv III" IIIV “I‘ll"! [I‘d-luv VII U IIIH ‘l'l'll ll. .II'I-"l I. ‘- 1| llllllJl 284| lillJlllJllll llllll 2837 2833 llllllllllllllllllll lllllllllllllll 2829 lllllllllllllllllllll IO 0 74 , II ”I I II I ‘ I I l / F l I IIVIAI IA I /II \Jll II I II I I ’I I" II II I ‘I, I I J I” III | I I II II: I I II 4 III ' I H I l I L ‘r‘ I I I ‘H I I I I V I I’d Ij ‘ I \ A I I I I I ,. In I I IJI I I I .I I ’II‘I'I I I I IIIl M II I” H" \a I I' ‘I \I r \ W U ‘\ \ l\\ l \ \JWV‘\J/\ A" I WAN. W“ N fi-va-r 'A‘J 11111!llljjlllllllllillljllllJlJllLJlJJlllllllllJllJlllll lLJJllLJJ llllllllllll|lllL1llJJllllllllllllllllllllll|ILLIIIIJJJII|I4JIIIIJll 284i 2845 2849 2853 l5 IO q JJH.“ fl [—1 l 5 v n I ,“I l ‘ l I ,\ n J /1 I I l / \I \ f4 I ll‘l l ./ \/ I III I \II I” I!" ’V' , u ‘ .g in: ‘1 I ‘1'“ f VII ‘ \ I \1|l bl] M ‘ ' w w M ’\ f \‘g 3 I f/4¢A/\/‘A'\/ d L' U MN“ " NMNMMM“~“MAm u ‘Mv\ -M—va VN—T‘v/F O IIIllllIlllllllll‘l III]iIllllllllllllllllllllllllllllllllllllllllllI IllIIIlllljllllllllllllllllll IJLJI|III lIIIIILIJIIIIIIIIIIIIIIIIIIII 2859 2863 285! 2855 76 ‘I-Iv \ II‘IIIIA. all|'llllll'lll|l|lll .lllll'll"-.. ‘I‘. rll'l ‘llal IIIII l'l..l.|.ll| Illlllululll (III .‘l'llll I. III, lull lllll"|y-.|ll!l.’ I.“ Ill-‘ IIIIIII 2863 l5 ID 2875 287| 286? 77 'lll ll'l'l'l Illl I I." ‘\ ul‘ 'IIII‘II I‘III lllll! IIIIIII \i'll ll'll] Illllll l 2887 IJJJllllllllll lllllllllllllll lllllllLJlllllllllll llllllllllll i 1 IllllljllJJLllllllvl lJLllllllIlllll lllllllllllllllll llll {5 IO 0 2883 2879 2875 78 H H H Ill“ Illllllllhv I lllllll .ll‘llllll'llu Il'lllllA lllllll llllll I’ll II-Illl lllllllalv AIIIIIIIIIIII I'll- "Illllll'l'l I IIIHIIHHJ 'lll'llllul‘lI-l Ill|i||| h! 'l'll'l'llllllll'l'I' ln'l llllllllll IIII'IN. ' / AHHHH IIIII llllll IIIIIIII‘IIIL \‘HU A lill‘llll Al I "J IlllllllllllIIllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllllIIIIIIJ llllllllllllllllllllllllllllllllllllllllllllllll | H H H HUIIIIHV 20 20